Анализ страницы https://catboost.ai/docs
Основное Готовность: 100%
Домен
catboost.ai
Состояние доменного имени
?
Проверяем корректность доменного имени и наличие технических проблем на уровне домена.
Домен второго уровня идеален для продвижения.
Отличный запоминающийся домен.
Ответ сервера
200 Успешный ответ
HTTP-код ответа и цепочка редиректов
?
Код 200 — страница доступна. Коды 3xx — редиректы (цепочки замедляют загрузку и размывают ссылочный вес). Коды 4xx/5xx — ошибки, поисковик не сможет проиндексировать страницу.
Сервер настроен корректно.
Цепочка редиректов:
https://catboost.ai/docs
302 Found
https://catboost.ai/docs/en/
200 OK
Безопасность
Сайт безопасен
Использование HTTPS и SSL-сертификат
?
HTTPS — обязательный стандарт. Google и Яндекс отдают предпочтение защищённым сайтам. Отсутствие SSL или просроченный сертификат ведут к предупреждениям в браузере и снижению позиций.
Не настроен HSTS (Strict-Transport-Security) — рекомендуется включить.
На сайте работает защищенный протокол ssl и сайт открывается по https.
Ssl-сертификат действителен до 13.11.2026 23:59:59.
Поздравляем! Сайт не содержится в реестре РКН.
Кодировка
utf-8
Кодировка символов страницы
?
Стандарт — UTF-8. Неправильная кодировка вызывает нечитаемые символы и мешает поисковику корректно распознать текст страницы.
Указана кодировка на странице utf-8.
Язык
en
Атрибут lang в HTML-теге
?
Атрибут lang (<html lang="ru">) сообщает поисковикам и браузерам, на каком языке написана страница. Помогает при ранжировании в региональном поиске.
Язык документа указан явно: en.
Скорость загрузки
~0,10сек
Время отклика сервера (TTFB)
?
Time To First Byte — время до получения первого байта от сервера. Норма до 200 мс. Медленный отклик ухудшает пользовательский опыт и ранжирование: Яндекс и Google учитывают скорость страниц.
Скорость загрузки сайта 0,10сек оптимальна.
Объем документа
124Кб
Размер HTML-кода страницы
?
Слишком большой HTML замедляет парсинг браузером и сканирование поисковым роботом. Рекомендуется не более 200 Кб.
Объем html-документа 124Кб оптимален.
Структура html-документа корректна.
Ресурсы
Ресурсы: 10
Внешние ресурсы страницы (CSS, JS, изображения)
?
Количество и тип подключённых ресурсов влияют на скорость загрузки. Большое число запросов увеличивает время рендеринга страницы.
Кол-во файлов ресурсов 10 достаточно.
Показать полный список ресурсов
| Тип | Название | Значение |
|---|---|---|
| stylesheet | https://storage.yandexcloud.net/docs-external/styles/katex/0.16.9/katex.min.css | |
| stylesheet | text/css | https://yastatic.net/s3/cloud/docs-viewer-external/static/freeze/en/_NaiOUfiZL6LLJMaTxO02/vendor.6d2b0128.css |
| stylesheet | text/css | https://yastatic.net/s3/cloud/docs-viewer-external/static/freeze/en/_NaiOUfiZL6LLJMaTxO02/common.c98427e6.css |
| js | https://yastatic.net/s3/cloud/docs-viewer-external/static/freeze/en/_NaiOUfiZL6LLJMaTxO02/runtime.47a25c03.js | |
| js | https://yastatic.net/s3/cloud/docs-viewer-external/static/freeze/en/_NaiOUfiZL6LLJMaTxO02/vendor.6d2b0128.js | |
| js | https://yastatic.net/s3/cloud/docs-viewer-external/static/freeze/en/_NaiOUfiZL6LLJMaTxO02/common.c98427e6.js | |
| js | https://yastatic.net/s3/cloud/docs-viewer-external/static/freeze/en/_NaiOUfiZL6LLJMaTxO02/pages/_app.js | |
| js | https://yastatic.net/s3/cloud/docs-viewer-external/static/freeze/en/_NaiOUfiZL6LLJMaTxO02/pages/_error.js | |
| js | https://yastatic.net/s3/cloud/docs-viewer-external/static/freeze/en/_NaiOUfiZL6LLJMaTxO02/pages/_doc.js | |
| js | https://yastatic.net/s3/cloud/docs-viewer-external/static/freeze/en/_NaiOUfiZL6LLJMaTxO02/init.de2c056c.js |
Серверные заголовки
Кол-во: 10
HTTP-заголовки ответа сервера
?
Заголовки сервера передают браузеру и поисковику служебную информацию: кеширование, безопасность (CSP, HSTS), сжатие (gzip). Правильная настройка ускоряет загрузку и повышает защищённость.
Найдены серверные заголовки 10шт. Подробнее про серверные заголовки.
Показать полный список серверных заголовков
| Ключ | Значение |
|---|---|
| Content-Security-Policy | frame-ancestors webvisor.com *.webvisor.com metrika.yandex.ru https://mc.yandex.ru; |
| Link | <https://catboost.ai/docs/en/index.yaml>; rel="alternate"; type="application/yaml"; title="Yaml version", <https://catboost.ai/docs/en/llms.txt>; rel="alternate"; type="text/markdown"; title="llms.txt" |
| x-trace-id | 1d89ebb4bd290cb661a3254655a98651 |
| X-XSS-Protection | 1; mode=block |
| Date | Fri, 21 Aug 2026 21:10:09 GMT |
| Cache-Control | no-store, must-revalidate |
| Surrogate-Control | no-store |
| X-Request-ID | 1787346609490532-12178864152574090066 |
| Set-Cookie | pi=6S6jpijkWS/4vkJZjyYDErgHDe1M9eQOj6N1udOCK65DX8f1fAKz0IayRbTa6m7vRXOKHebgGfZ3bdy7+65rlv9Lg8g=; Expires=Sun, 20-Aug-2028 21:10:09 GMT; Domain=.yandex.ru; Path=/; Secure; HttpOnly; SameSite=None; Partitioned |
| X-Content-Type-Options | nosniff |
CMS
Diplodoc Platform v5.54.5
Система управления сайтом (движок)
?
CMS — это движок, на котором работает сайт (WordPress, 1C-Bitrix, Tilda и др.). Знание CMS помогает понять возможности SEO-оптимизации и подобрать подходящие инструменты. «Не определена» — вероятно, самописный сайт или нестандартная сборка.
В мета-теге generator указано: Diplodoc Platform v5.54.5.
Веб-сервер
Не определён
Программное обеспечение сервера
?
Веб-сервер — это ПО, которое отдаёт страницы посетителям (nginx, Apache, IIS, LiteSpeed и др.). Определяется по серверным заголовкам ответа (Server, X-Powered-By и т.п.). «Не определён» — сервер намеренно скрывает эти заголовки, это нормальная практика безопасности.
Веб-сервер не определён — заголовок Server скрыт. Это не ошибка: часто так настраивают из соображений безопасности.
Мета-теги Готовность: 63%
Title
CatBoost
Заголовок страницы в браузере и поисковой выдаче
?
Title — главный SEO-заголовок страницы. Влияет на CTR в поиске и ранжирование. Оптимальная длина: 50–70 символов. Ключевые слова — ближе к началу.
Необходимо увеличить число символов в title (текущее значение мало: 8, минимум: 25, оптимально: от 40 до 45)
Дублей словоформ в title не найдено.
Description
CatBoost is a machine learning algorithm that uses gradient boosting on decision trees. It is available as an open source library.
Описание страницы в поисковой выдаче (сниппет)
?
Meta Description — текст под заголовком в выдаче. Напрямую на позиции не влияет, но влияет на CTR. Оптимальная длина: 120–160 символов.
Число символов в description 130 оптимально (норма: от 120 до 130).
Keywords
Список ключевых слов страницы (устаревший тег)
?
Meta Keywords не учитывается Яндексом и Google для ранжирования с 2009–2012 годов. Заполнение не обязательно, но не вредит. Конкурент может использовать содержимое для анализа.
Установите мета-тег keywords!
Канонический Url
en/
Указывает поисковику основную версию страницы
?
Canonical (rel=canonical) предотвращает проблему дублей страниц. Должен точно совпадать с URL проверяемой страницы. Неправильный canonical может передать ссылочный вес на другую страницу.
Домен в каноническом Url не совпадает!
Robots
Ошибок нет
Директивы для поисковых роботов на уровне страницы
?
Meta Robots управляет индексацией конкретной страницы: index/noindex — индексировать ли, follow/nofollow — следовать ли по ссылкам. Noindex полностью исключает страницу из поиска.
Meta-тег robots не указан. Страница свободна для индексации.
Адаптивность
width=device-width, initial-scale=1.0
Настройка масштабирования на мобильных устройствах
?
Тег viewport (<meta name="viewport">) сообщает браузеру, как масштабировать страницу на мобильных. Стандарт: width=device-width, initial-scale=1. Отсутствие — признак отсутствия мобильной версии.
Meta-тег viewport со значением-константой width=device-width задаёт ширину страницы в соответствии с размером экрана.
Meta-тег viewport со значением initial-scale=1.0 определяет масштаб 1:1, т.е. «не масштабировать».
Разметка OpenGraph
Кол-во: 6
Мета-теги для красивых превью в соцсетях
?
OpenGraph (og:title, og:description, og:image) управляет тем, как страница выглядит при репосте в социальных сетях и мессенджерах. Отсутствие OG-тегов — невзрачный превью при шеринге.
Разметка OpenGraph задана. Страница оптимизирована под социальные сети.
Показать полный список og мета-тегов
| Тип | Значение |
|---|---|
| og:type | article |
| og:url | en/ |
| og:title | CatBoost |
| og:description | CatBoost is a machine learning algorithm that uses gradient boosting on decision trees. It is available as an open source library. |
| og:image | https://catboost-opensource.s3.yandex.net/logo_white_for_dark_theme.svg |
| og:locale | en_EN |
Все мета-теги
Кол-во: 17
Полный список мета-тегов страницы
?
Таблица всех meta-тегов, включая нестандартные. Позволяет найти опечатки, дубли и лишние теги.
Найдены мета-теги 17шт. Мета-теги не видимы для человека и предназначены для обмена информацией между веб-страницей и поисковыми системами, браузерами и другими веб-службами. С ними роботы 🤖 и устройства ведут себя более ожидаемо.
Показать полный список мета-тегов
| Тип | Название | Значение |
|---|---|---|
| name | viewport | width=device-width, initial-scale=1.0 |
| name | langs | en |
| name | lang | en |
| name | generator | Diplodoc Platform v5.54.5 |
| name | twitter:card | summary_large_image |
| name | twitter:title | CatBoost |
| name | twitter:description | CatBoost is a machine learning algorithm that uses gradient boosting on decision trees. It is available as an open source library. |
| name | twitter:image | https://catboost-opensource.s3.yandex.net/logo_white_for_dark_theme.svg |
| name | description | CatBoost is a machine learning algorithm that uses gradient boosting on decision trees. It is available as an open source library. |
| property | og:type | article |
| property | og:url | en/ |
| property | og:title | CatBoost |
| property | og:description | CatBoost is a machine learning algorithm that uses gradient boosting on decision trees. It is available as an open source library. |
| property | og:image | https://catboost-opensource.s3.yandex.net/logo_white_for_dark_theme.svg |
| property | og:locale | en_EN |
| property | share:content_icon | catboost |
| property | share:title | CatBoost |
Оптимизация Готовность: 55%
Структура
Ошибок нет
Семантические HTML-элементы страницы
?
Проверяет наличие основных структурных элементов: nav, header, footer, main. Корректная семантическая структура помогает поисковику понять архитектуру страницы.
Структура документа корректна (теги <html> и <body> присутствуют по одному на документ).
Контент
Есть ошибки
Объём и качество текстового содержимого
?
Анализирует объём полезного текста на странице. Слишком мало — страница может считаться малополезной. Слишком много — ухудшается читаемость и восприятие.
Слова из title 1 встречаются в тексте редко. Добавьте в контент страницы слова из тега <title>!
Абзацев с текстом 0 слишком мало. Добавьте больше абзацев с текстом (тег <p>)!
Среднее число слов в абзаце 0 слишком мало. Сделайте контент более читаемым!
Кол-во слов 153 слишком мало. Добавьте больше текста (минимум 400 слов)!
Кол-во знаков контента 1201 на странице оптимально.
Заголовки
Ошибок нет
Иерархия заголовков H1–H6
?
H1 должен быть один и содержать ключевой запрос. H2–H6 описывают подразделы. Пропуск уровней (H1 → H3) и несколько H1 — типичные ошибки, снижающие понятность страницы для поисковика.
На странице присутствуют заголовки <h1> 1. Это прекрасно.
На странице присутствуют заголовки <h2> 8. Это хорошо.
Тошнота
2,24
Насколько одно слово доминирует в тексте
?
Классическая тошнота = √(частота самого повторяющегося слова). Норма до 7–8: текст воспринимается естественно. Выше — поисковик может счесть страницу переспамленной.
Тошнота страницы в пределах нормы 3.
Академич. тошнота
31,37%
Насколько текст перенасыщен ключевыми словами
?
Академическая тошнота = (частота слова / общее количество слов) × 100%. Показывает долю конкретного слова в тексте. Норма 5–15%.
Академическая тошнота превышает норму 5-15%. Измените текст страницы!
Семантическое ядро
20
Наиболее часто встречающиеся слова на странице
?
Топ слов по частоте использования. Показывает, какие слова доминируют в тексте с точки зрения поисковика.
Контент страницы содержит осмысленный текст и слова.
Показать список слов
| Слово | Кол-во | Частота |
|---|---|---|
| training | 5 | 3,27% |
| metrics | 5 | 3,27% |
| installation | 4 | 2,61% |
| python | 4 | 2,61% |
| package | 4 | 2,61% |
| catboost | 4 | 2,61% |
| features | 4 | 2,61% |
| applying | 3 | 1,96% |
| models | 3 | 1,96% |
| apache | 2 | 1,31% |
| command-line | 2 | 1,31% |
| version | 2 | 1,31% |
| source | 2 | 1,31% |
| analysis | 2 | 1,31% |
| visualization | 2 | 1,31% |
| algorithm | 2 | 1,31% |
| educational | 2 | 1,31% |
| materials | 2 | 1,31% |
| prediction | 2 | 1,31% |
| importances | 2 | 1,31% |
Индексация Готовность: 80%
Индексирование
Ошибок нет
Разрешено ли индексирование страницы
?
Проверяет, не закрыта ли страница от индексации через robots.txt, meta robots или X-Robots-Tag. Страница, закрытая от индексации, не появится в поисковой выдаче.
Анкоров на странице 39 оптимально. Поисковые роботы обязательно проиндексируют сайт.
Robots.txt
Найден корректный robots.txt
Файл управления сканированием сайта роботами
?
Robots.txt указывает поисковым роботам, какие страницы сканировать, а какие — нет. Ошибки в файле могут случайно закрыть важные разделы от индексации.
Robots.txt настроен корректно. Размер файла: 75294 байт. Загружен за: 83мсек.
Проверяемая страница не запрещена в robots.txt.
Robots.txt доступен по постоянному адресу
Показать содержимое robots.txt
<!doctype html>
<html lang="en">
<head>
<title data-react-helmet="true">CatBoost - open-source gradient boosting library</title>
<meta data-react-helmet="true" name="description" content="CatBoost is an open-source gradient boosting on decision trees library with categorical features support out of the box, successor of the MatrixNet algorithm developed by Yandex."/><meta data-react-helmet="true" name="keywords" content="catboost, matrixnet, boosting, gradient boosting, gradient boosting on decision trees, decision trees, classification, regression, LightGBM, XGBoost"/><meta data-react-helmet="true" property="og:type" content="website"/><meta data-react-helmet="true" property="og:title" content="CatBoost - state-of-the-art open-source gradient boosting library with categorical features support"/><meta data-react-helmet="true" property="og:description" content="CatBoost - state-of-the-art open-source gradient boosting library with categorical features support, https://catboost.yandex/ #catboost"/><meta data-react-helmet="true" property="og:image" content="https://avatars.mds.yandex.net/get-bunker/56833/dba868860690e7fe8b68223bb3b749ed8a36fbce/orig"/><meta data-react-helmet="true" property="og:url" content="https://catboost.ai"/><meta data-react-helmet="true" name="twitter:card" content="summary_large_image"/><meta data-react-helmet="true" name="twitter:title" content="CatBoost - state-of-the-art open-source gradient boosting library with categorical features support"/><meta data-react-helmet="true" name="twitter:description" content="#CatBoost - state-of-the-art open-source gradient boosting library with categorical features support, "/><meta data-react-helmet="true" name="twitter:image" content="https://avatars.mds.yandex.net/get-bunker/56833/dba868860690e7fe8b68223bb3b749ed8a36fbce/orig"/><meta data-react-helmet="true" name="viewport" content="width=device-width, maximum-scale=1.0, minimal-ui"/><meta data-react-helmet="true" name="format-detection" content="telephone=no"/>
<link data-react-helmet="true" rel="icon" type="image/png" href="https://avatars.mds.yandex.net/get-bunker/49769/c4b06f2b169c7596da4143d5f951d925b17c413e/orig"/><link data-react-helmet="true" rel="apple-touch-icon" href="https://avatars.mds.yandex.net/get-bunker/49769/a74e0923e4e58febf68787226ef24769aea5a2e5/orig"/><link data-react-helmet="true" rel="apple-touch-icon" sizes="120x120" href="https://avatars.mds.yandex.net/get-bunker/49769/96a4132b5fccfbe6a8a18a71279626bcabe2ba62/orig"/>
<style>body{padding:0}*,body{margin:0}body{min-width:948px;font-weight:400;color:#333;-webkit-font-smoothing:antialiased;-moz-osx-font-smoothing:grayscale;-webkit-text-size-ajust:none}body,button,input{font-family:YS Text,sans-serif}button,input{outline:none}a{text-decoration:none;color:#1a7aff}a:hover{color:#5fa2ff}input{-webkit-appearance:none}input:invalid{box-shadow:none}button{cursor:pointer}.column{width:276px}.row{display:-ms-flexbox;display:flex;-ms-flex-pack:justify;justify-content:space-between}@font-face{font-family:YS Text;font-weight:300;font-style:normal;font-stretch:normal;src:url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/a8d6e75f9531791307faf1cd90a7ac81.woff2) format("woff2"),url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/8c00b3c529861eff8f9ff61cff55522e.woff) format("woff")}@font-face{font-family:YS Text;font-weight:300;font-style:italic;font-stretch:normal;src:url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/018c2a0d68482b132b28c0e2e53ac5b0.woff2) format("woff2"),url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/67af0a6d95970d560a123f4dd7b12785.woff) format("woff")}.y-font_ys-text-light,.y-font_ys-text-lightitalic{font-family:YS Text,sans-serif;font-weight:300}.y-font_ys-text-lightitalic{font-style:italic}@font-face{font-family:YS Text;font-weight:400;font-style:normal;font-stretch:normal;src:url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/48952179855a7c9189bff82e349bcbfb.woff2) format("woff2"),url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/51a98bd1d7ce72abb481e75c57bf9b2a.woff) format("woff")}@font-face{font-family:YS Text;font-weight:400;font-style:italic;font-stretch:normal;src:url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/00880048bf1266d084a1c56a19a7211a.woff2) format("woff2"),url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/50e4f4c020a08d6ea9d87b3ea980b9c5.woff) format("woff")}.y-font_ys-text-regular,.y-font_ys-text-regularitalic{font-family:YS Text,sans-serif;font-weight:400}.y-font_ys-text-regularitalic{font-style:italic}@font-face{font-family:YS Text;font-weight:500;font-style:normal;font-stretch:normal;src:url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/02056ad2da1a869187aadc98cd0e7996.woff2) format("woff2"),url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/09559949bfdba9f8ce8f92aecc6e6b0e.woff) format("woff")}@font-face{font-family:YS Text;font-weight:500;font-style:italic;font-stretch:normal;src:url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/c9d947ada94384220117596d3b177296.woff2) format("woff2"),url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/5466e6baa8e4f84b6fadbfb5e9ee9628.woff) format("woff")}.y-font_ys-text-medium,.y-font_ys-text-mediumitalic{font-family:YS Text,sans-serif;font-weight:500}.y-font_ys-text-mediumitalic{font-style:italic}@font-face{font-family:YS Text;font-weight:600;font-style:normal;font-stretch:normal;src:url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/9044f2fdff676bdd35bf8e746618c4eb.woff2) format("woff2"),url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/bb62276bed2683f8818e1bdca0746b2b.woff) format("woff")}@font-face{font-family:YS Text;font-weight:600;font-style:italic;font-stretch:normal;src:url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/8c0b02b87a517bfee00fb0930f72f497.woff2) format("woff2"),url(//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/dba7e4a90cff3890e1ad7b414ec142f3.woff) format("woff")}.y-font_ys-text-bold,.y-font_ys-text-bolditalic{font-family:YS Text,sans-serif;font-weight:600}.y-font_ys-text-bolditalic{font-style:italic}.container{width:868px;min-width:868px;margin:0 auto}.navbar{position:fixed;z-index:10;top:0;display:-ms-flexbox;display:flex;width:100%;min-width:868px;height:70px;font-size:16px;font-weight:400;background-color:#fff;transition:box-shadow .25s cubic-bezier(.17,.84,.44,1);-ms-flex-pack:justify;justify-content:space-between}.navbar__links-wrapper{width:100%}.navbar__links-wrapper,.navbar__social-links-container{display:-ms-flexbox;display:flex;-ms-flex-direction:row;flex-direction:row;-ms-flex-pack:center;justify-content:center;-ms-flex-align:center;align-items:center}.navbar__social-links-container{position:absolute;top:0;right:calc(50% - 868px / 2);height:100%;padding:0;list-style-type:none;-ms-flex-negative:0;flex-shrink:0}.navbar__social-link-li{height:100%}.navbar__social-link{display:block;width:23px;height:100%;margin:0 12px;opacity:1;background-repeat:no-repeat;background-position:50%;background-size:23px 19px;transition:opacity .4s linear}.navbar__social-link:hover{opacity:.8}.navbar__social-link_twitter{background-image:url("data:image/svg+xml,%3Csvg width='23' height='19' viewBox='0 0 23 19' xmlns='http://www.w3.org/2000/svg'%3E%3Cpath d='M.19 16.87a13.41 13.41 0 0 0 12.92.9 13.04 13.04 0 0 0 5.68-5.23 14 14 0 0 0 1.95-7v-.4a9.83 9.83 0 0 0 2.19-2.33.42.42 0 0 0-.12-.58.4.4 0 0 0-.4-.04c-.32.14-.64.26-.97.36.42-.49.72-1.06.91-1.67a.41.41 0 0 0-.27-.52.43.43 0 0 0-.34.04 8.6 8.6 0 0 1-2.55 1 4.93 4.93 0 0 0-7.02.13 5 5 0 0 0-1.35 4.07A12.43 12.43 0 0 1 2.27 1a.4.4 0 0 0-.58-.06l-.1.12a5 5 0 0 0 .42 5.62 3.95 3.95 0 0 1-.51-.24.42.42 0 0 0-.57.17.35.35 0 0 0-.04.19v.07c0 1.69.85 3.27 2.26 4.2l-.18-.04a.41.41 0 0 0-.48.33.38.38 0 0 0 .01.2 4.97 4.97 0 0 0 3.53 3.33A8.66 8.66 0 0 1 .46 16.1a.41.41 0 0 0-.46.37c-.01.15.06.3.19.4z' fill-rule='nonzero' fill='%23858585'/%3E%3C/svg%3E")}.navbar__social-link_telegram{background-image:url("data:image/svg+xml,%3Csvg width='23' height='19' viewBox='0 0 23 19' xmlns='http://www.w3.org/2000/svg'%3E%3Cpath d='M21.52.07L.5 8.11a.75.75 0 0 0 .06 1.42l5.35 1.55 1.99 6.2a.93.93 0 0 0 1.53.37l2.77-2.77 5.42 3.9c.66.48 1.6.13 1.77-.66l3.6-16.84c.17-.82-.66-1.51-1.46-1.2zM18.9 3.81l-9.77 8.47c-.1.09-.16.2-.18.33l-.37 3.28c-.02.1-.17.12-.2.02L6.82 11a.51.51 0 0 1 .23-.58L18.58 3.4c.27-.16.54.2.3.4z' fill='%23858585' fill-rule='nonzero'/%3E%3C/svg%3E")}.navbar__links{display:-ms-flexbox;display:flex;height:100%}.navbar__link{height:100%;padding:0 16px;cursor:pointer;vertical-align:middle;white-space:nowrap;text-decoration:none;line-height:70px;color:#707070;transition:color .2s ease-out}.navbar__link:hover{color:#252525}.navbar__logo{position:absolute;top:0;left:0;display:-ms-flexbox;display:flex;height:100%;padding-right:16px}.navbar__arrow{position:relative;min-width:60px;height:100%;background-color:#fc0;background-repeat:no-repeat;background-position:100%}.navbar__arrow:after{position:absolute;right:-20px;display:block;box-sizing:border-box;width:0;height:0;content:"";border-top:35px solid transparent;border-bottom:35px solid transparent;border-left:20px solid #fc0}.navbar__title{min-width:143px;height:100%;background-repeat:no-repeat;background-position:100%}.navbar_mobile{position:relative}.navbar_scrolling{box-shadow:0 1px 0 rgba(12,13,14,.1),0 1px 3px rgba(12,13,14,.1),0 4px 20px rgba(12,13,14,.04),0 1px 1px rgba(12,13,14,.03)}@media (max-width:1320px){.navbar__links-wrapper{margin-right:calc(50% - 868px / 2);-ms-flex-pack:justify;justify-content:space-between}.navbar__logo,.navbar__social-links-container{position:static}}@media (max-width:1024px){.navbar__title{display:none}}.head{box-sizing:border-box;height:325px;margin-bottom:50px;padding-top:114px;background-size:cover}.head_mobile{height:255px;padding-top:44px}.head__text{margin-top:0;margin-bottom:20px;font-size:32px;line-height:40px;color:#333}.head__button{display:inline-block;padding:0 35px;vertical-align:middle;text-decoration:none;font-size:20px;font-weight:700;line-height:50px;color:#1a7aff;border:1px solid #d7d7d7;border-radius:4px;background-color:#f7f9fa;transition:border .2s linear,background .3s linear}.head__button:hover{border:1px solid #1a7aff}.head__button:not(:last-child){margin-right:16px}.about-video{position:relative;width:868px;height:490px;background-color:#000}.about-video:before{position:absolute;z-index:1;top:0;left:-1px;display:block;width:5px;height:100%;content:"";background-color:#ed4646}.about-video__iframe{width:100%;height:100%}.feature{position:relative;display:-ms-flexbox;display:flex;-ms-flex-direction:column;flex-direction:column;box-sizing:border-box;height:0;min-height:250px;margin-bottom:30px;color:#333}.feature__picture{width:90px;transition:.3s cubic-bezier(.4,0,.2,1)}.feature:hover .feature__picture{transform:scale(1.44)}.feature__header{display:-ms-flexbox;display:flex;height:90px;font-size:24px;font-weight:700;line-height:20px;border-top:1px solid #ddd;border-right:1px solid #ddd;border-top-right-radius:5px;background-color:#f6f8fa}.feature__number{width:90px;text-align:center;-ms-flex-pack:center;justify-content:center}.feature__number,.feature__title{display:-ms-flexbox;display:flex;-ms-flex-align:center;align-items:center}.feature__title{overflow:hidden;padding-left:39px;background-repeat:no-repeat;background-position:0 100%;background-size:30px;-ms-flex-positive:1;flex-grow:1}.feature__spacer{position:absolute;z-index:1;top:0;left:0;width:5px;height:100%}.feature__text{padding-top:20px;padding-right:80px;padding-left:90px;font-size:20px;font-weight:400;line-height:30px;color:#333;border-right:1px solid #ddd;border-bottom:1px solid #ddd;border-bottom-right-radius:5px;-ms-flex-positive:1;flex-grow:1}.title{margin-bottom:28px;font-size:38px;font-weight:700;line-height:42px}.footer{margin-top:100px}.footer__contact-links-container{display:-ms-flexbox;display:flex;-ms-flex-direction:column;flex-direction:column;padding:20px 0;border-top:1px solid #d8d8d8;border-bottom:1px solid #d8d8d8;-ms-flex-align:start;align-items:flex-start}.footer__copyright{display:-ms-flexbox;display:flex;box-sizing:border-box;height:65px;padding-top:19px;padding-bottom:22px;font-size:15px;font-weight:400;color:#999;-ms-flex-pack:justify;justify-content:space-between;-ms-flex-align:center;align-items:center}.footer__link-item{display:block;font-size:16px;line-height:26px}.footer__link-item:before{margin-right:4px;content:"\2014"}.last-news{margin-bottom:100px}.last-news__button-wrapper{display:-ms-flexbox;display:flex;margin-top:50px}.last-news__button{margin:auto;padding:12px 39px 14px;text-align:center;text-decoration:none;font-size:20px;font-weight:600;color:#1a7aff;border:1px solid #d6d6d6;border-radius:4px;background:#f6f8fa;transition:border .2s linear}.last-news__button:hover{border:1px solid #1a7aff}.last-news__loading-failed{color:#d0021b}.news-item__header{position:relative;box-sizing:border-box;padding-top:20px;padding-bottom:48px;border-top:1px solid #d8d8d8;border-bottom:1px solid #d8d8d8}.news-item__title{display:block;text-decoration:none;font-size:24px;font-weight:700;line-height:30px;color:#1a7aff;transition:opacity .3s ease-out}.news-item__title:hover{opacity:.5}.news-item__date{position:absolute;bottom:10px;font-size:13px;opacity:.5}.news-item__date,.news-item__text{font-weight:400;line-height:25px;color:#333}.news-item__text{padding-top:20px;font-size:15px}.news-item__text p{margin:0}.newsletter{position:absolute;width:100%;height:100%}.newsletter__group{position:relative;width:100%;height:240px;margin-top:100px;margin-bottom:100px;border-top:1px solid #d8d8d8;border-bottom:1px solid #d8d8d8}.newsletter__screen-exit{opacity:1}.newsletter__screen-exit-active{opacity:0;transition:opacity .5s ease-in}.newsletter__screen-enter{opacity:0}.newsletter__screen-enter-active{z-index:2;opacity:1;transition:opacity .5s ease-out .5s}.newsletter__title{margin:32px 0 8px;text-align:center;font-size:24px}.newsletter__description{margin:20px 0;padding:0 120px;text-align:center;font-size:15px}.newsletter__form-wrapper{position:absolute;top:140px;width:100%}.newsletter__form{display:-ms-flexbox;display:flex;height:50px}.newsletter__input{min-width:300px;margin-left:auto;padding:0 16px;font-size:20px;font-weight:700;color:#333;border:1px solid #d8d8d8;border-right:0;border-radius:4px 0 0 4px;transition:border .2s ease-in-out,background-color .2s ease-in-out}.newsletter__input:focus,.newsletter__input:hover{border-color:#1a7aff}.newsletter__input:disabled{border-color:#d8d8d8;background-color:#fff}.newsletter__input::-webkit-input-placeholder{color:#d8d8d8}.newsletter__input:-ms-input-placeholder{color:#d8d8d8}.newsletter__input::placeholder{color:#d8d8d8}.newsletter__input_invalid{border-color:#d0021b;background-color:rgba(208,2,27,.06)}.newsletter__input_invalid:focus,.newsletter__input_invalid:hover{border-color:#d0021b}.newsletter__input_invalid::-webkit-input-placeholder{color:rgba(208,2,27,.41)}.newsletter__input_invalid:-ms-input-placeholder{color:rgba(208,2,27,.41)}.newsletter__input_invalid::placeholder{color:rgba(208,2,27,.41)}.newsletter__button{margin-right:auto;padding:0 40px;font-size:20px;font-weight:700;color:#fff;border:none;border-radius:0 4px 4px 0;background-color:#1a7aff;transition:background-color .2s ease-in-out}.newsletter__button:hover{background-color:#5fa2ff}.newsletter__button:disabled{background-color:rgba(26,122,255,.4)}.newsletter__terms-conditions{margin-top:10px;text-align:center;font-size:15px;line-height:28px;color:#9b9b9b}.newsletter__terms-conditions a{text-decoration:none;color:#4a90e2}.newsletter__terms-conditions a:hover{color:#1a7aff}.newsletter__captcha-wrapper{display:-ms-flexbox;display:flex;margin:8px 0 14px;-ms-flex-align:center;align-items:center}.newsletter__captcha{width:200px;height:60px;margin-left:auto;background-repeat:no-repeat;background-position:50%;background-size:200px 60px}.newsletter__captcha-button{display:-ms-flexbox;display:flex;height:32px;margin-left:30px;padding:7px 0;font-size:15px;color:#1a7aff;border:none;background-color:transparent;transition:color .2s ease-in-out}.newsletter__captcha-button:hover{color:rgba(26,122,255,.7)}.newsletter__captcha-button:last-child{margin-right:auto}.newsletter__icon{width:18px;height:18px;margin-right:4px;background-repeat:no-repeat;background-position:50%}.newsletter__icon_refresh{height:17px;background-size:14px 17px}.newsletter__icon_voice{background-size:17px 16px}.newsletter__error{margin-top:15px;text-align:center;font-size:15px;color:#d0021b}.newsletter_success{display:table}.newsletter__success{display:table-cell;height:100%;text-align:center;vertical-align:middle;font-size:24px;font-weight:700;color:#1a7aff}.table-chart-controls{flex-direction:row;margin-bottom:30px}.table-chart-controls,.table-chart-controls__group{display:-ms-flexbox;display:flex;-ms-flex-direction:row}.table-chart-controls__group{flex-direction:row;-ms-flex:0 1 auto;flex:0 1 auto;-ms-flex-align:center;align-items:center;-ms-flex-pack:justify;justify-content:space-between}.table-chart-controls__label{display:-ms-flexbox;display:flex;height:50px;margin:0 52px 0 0;cursor:pointer;font-size:20px;font-weight:700;-ms-flex-align:center;align-items:center}.table-chart-controls__button{box-sizing:border-box;width:36px;height:36px;margin:0 16px 0 0;cursor:pointer;-webkit-user-select:none;-moz-user-select:none;-ms-user-select:none;user-select:none;text-align:center;color:#1a7aff;border:3px solid #1a7aff;border-radius:4px;outline:none;background:transparent no-repeat 50%;background-size:21px;transition:border .2s cubic-bezier(.4,0,.2,1),background-color .2s cubic-bezier(.4,0,.2,1)}.table-chart-controls__button:hover{border-color:#5fa2ff}.table-chart-controls__button_active{color:#fff;border:3px solid transparent;background-color:#1a7aff;background-size:21px}.table-chart-controls__button_active:hover{background-color:#5fa2ff}.table-chart-controls__button_title{width:432px;margin-left:0;cursor:default;color:#4a4a4a}.table-chart-controls__button_title:hover{border:1px solid #d7d7d7}.table-chart-controls__radio-label{height:50px;margin:0 48px 0 0;padding:7px 24px 7px 0;cursor:pointer;font-size:20px;font-weight:700;border-radius:4px;transition:color .2s cubic-bezier(.4,0,.2,1),background-color .2s cubic-bezier(.4,0,.2,1);-ms-flex-align:center;align-items:center}.table-chart-controls__radio-label_active{color:#fff;background-color:#1a7aff}.table-chart-controls__radio{position:absolute;box-sizing:border-box;width:36px;height:36px;margin:0 16px 0 0;cursor:pointer;-webkit-user-select:none;-moz-user-select:none;-ms-user-select:none;user-select:none;color:#1a7aff;border:3px solid #1a7aff;border-radius:18px;outline:none;background:transparent;transition:border .2s cubic-bezier(.4,0,.2,1)}.table-chart-controls__radio:hover{border-color:#5fa2ff}.table-chart-controls__radio_active{visibility:hidden;width:0;margin:0;padding:0;border-width:0;background-color:#1a7aff}.table-chart-controls__radio-text{position:relative;margin-left:52px;padding-top:6px;padding-bottom:6px;transition:margin-left .2s cubic-bezier(.4,0,.2,1)}.table-chart-controls__radio-text_active{margin-left:24px}.table-chart-header{margin-top:54px;transition:transform .5s ease-in-out}.table-chart-header_subheader{transform:translateY(0)}.table-chart-header__row{position:relative;display:-ms-flexbox;display:flex;-ms-flex-direction:row;flex-direction:row;padding-left:175px}.table-chart-header__row_line:before{position:absolute;top:0;left:0;display:block;width:100%;height:1px;content:"";background-color:#d8d8d8}.table-chart-header__main-item{margin-bottom:32px;padding-left:4px;font-size:20px;font-weight:700;-ms-flex:1 1 0px;flex:1 1 0}.table-chart-header__sub-item{display:block;overflow:hidden;padding:11px 0 11px 4px;font-size:15px;font-weight:700;color:#9b9b9b;transition:transform .4s ease-in-out,opacity .4s linear;-ms-flex:1 1 0px;flex:1 1 0}.table-chart-header__sub-item_thin{letter-spacing:-1.12px}.table-chart-header__sub-item_hidden{opacity:0}.table-chart-header__sub-item_translate{transform:translateX(-87px)}.table-chart-row{position:relative;display:-ms-flexbox;display:flex;overflow:hidden;-ms-flex-direction:row;flex-direction:row;height:80px}.table-chart-row:before{position:absolute;z-index:2;top:0;display:block;width:100%;height:1px;content:"";background-color:#d8d8d8}.table-chart-row__dataset{position:relative;width:175px;padding-top:5px;padding-left:17px;color:#000;border-top:1px solid #d8d8d8}.table-chart-row__item{display:inline-block;text-decoration:none;font-size:15px;font-weight:700;color:#1a7aff;transition:opacity .3s linear}.table-chart-row__item:hover{opacity:.4}.table-chart-row__item:before{position:absolute;top:6px;left:0;display:block;width:17px;height:18px;content:"";background-image:url("data:image/svg+xml,%3Csvg width='12' height='12' viewBox='0 0 12 12' xmlns='http://www.w3.org/2000/svg'%3E%3Cpath d='M5.38 0l1.95 1.95L4.7 4.59l2.64 2.65 2.65-2.65 1.95 1.95V0H5.38zM0 0v12h12V9.6H2.4V0H0z' fill='%231A7AFF' fill-rule='evenodd'/%3E%3C/svg%3E");background-repeat:no-repeat;background-position:0;background-size:12px 12px}.table-chart-row__item_plain{margin-left:-17px;color:#333}.table-chart-row__item_plain:before{content:none}.table-chart-row__item_plain:hover{opacity:1}.table-chart-cell{position:relative;box-sizing:border-box;transition:transform .4s ease-in-out,background .4s linear;-ms-flex:1 1 0px;flex:1 1 0}.table-chart-cell_pos_hide{transform:translateX(-87px)}.table-chart-cell_pos_wide{transform:translateX(87px)}.table-chart-cell__value{position:absolute;top:7px;left:4px;font-size:15px}.table-chart-cell__value_min{font-weight:700}.table-chart-cell__percentage{position:absolute;top:27px;left:4px;visibility:visible;font-size:15px;opacity:1;transition:opacity .5s ease 0s,visibility 0s linear 0s}.table-chart-cell__percentage:before{display:inline-block;content:"\2212"}.table-chart-cell__percentage_arrow_positive:before{content:"+"}.table-chart-cell__percentage_arrow_zero:before{display:none}.table-chart-cell__percentage_hidden{visibility:hidden;opacity:0;transition:opacity .5s ease 0s,visibility 0s linear .5s}.table-chart-cell__percentage-symbol{font-size:15px}.table-chart-cell__animate-enter{opacity:.01;transition:opacity .5s linear}.table-chart-cell__animate-enter-active{opacity:1}.table-chart-cell__animate-leave{opacity:1;transition:opacity .5s linear}.table-chart-cell__animate-leave-active{opacity:.01}.table-chart-legend{position:relative}.table-chart-legend a{text-decoration:none;color:#4a90e2}.table-chart-legend a:hover{color:#1a7aff}.table-chart-legend:before{display:block;width:100%;height:1px;margin-bottom:16px;content:"";background-color:#d8d8d8}.table-chart-legend__subtext{font-size:15px;line-height:17px;color:#9b9b9b}.table-chart *{box-sizing:border-box}.table-chart-grade{background-color:#fff}.table-chart-grade_level_0{font-weight:700;line-height:1.1}.table-chart-grade_level_0 .table-chart-cell__percentage{font-family:YS Regular Text,sans-serif;font-weight:400}.benchmarks{-webkit-tap-highlight-color:transparent}.benchmarks__tab-list{display:-ms-flexbox;display:flex;margin-top:28px;margin-bottom:42px;padding:0;border-bottom:1px solid #d8d8d8;-webkit-tap-highlight-color:black}.benchmarks__tab-item{margin-right:42px;padding:0 4px 8px;list-style:none;cursor:pointer;font-size:24px;font-weight:700;color:#9b9b9b;outline:none}.benchmarks__tab-item--selected{color:#333;border-bottom:12px solid #1a7aff}.benchmarks__tab-panel{-webkit-tap-highlight-color:black}.main-page__section{margin-bottom:100px}.main-page__paragraph{margin-top:0;margin-bottom:40px;padding-top:30px;font-size:32px;line-height:45px;color:#858585;border-top:1px solid #d8d8d8}.main-page__paragraph a{text-decoration:none;font-weight:700;color:#1a7aff;transition:opacity .3s ease-out}.main-page__paragraph a:hover{opacity:.5}.main-page__paragraph b{color:#333}.news-page-item{position:relative;display:-ms-flexbox;display:flex;-ms-flex-direction:column;flex-direction:column;box-sizing:border-box;width:424px;height:268px;margin-bottom:30px;padding:20px 21px 23px 18px;text-decoration:none;border:1px solid #ddd;border-radius:5px;-ms-flex:0 0 auto;flex:0 0 auto}.news-page-item:nth-child(odd){margin-right:10px}.news-page-item:nth-child(2n){margin-left:10px}.news-page-item__title{margin-bottom:14px;letter-spacing:.25px;font-size:24px;line-height:28px;color:#333}.news-page-item__date{letter-spacing:0;font-size:13px;line-height:15px;opacity:.5;color:#333}.news-page-item__preview{position:relative;overflow:hidden;max-height:110px;margin-top:11px;margin-bottom:37px;letter-spacing:0;font-size:15px;line-height:22px;color:#333}.news-page-item__preview p{margin:0}.news-page-item__preview_cut:after{position:absolute;right:0;bottom:0;width:35%;height:22px;content:"";background:linear-gradient(90deg,hsla(0,0%,100%,0),#fff 70%)}.news-page-item__read-more{position:absolute;z-index:1;bottom:22px;margin-top:auto;letter-spacing:0;font-size:15px;line-height:22px;color:#1a7aff}.news-page-load-more{display:table;box-sizing:border-box;margin:auto;padding:12px 39px 14px;cursor:pointer;text-align:center;font-size:20px;font-weight:600;color:#1a7aff;border:1px solid #d6d6d6;border-radius:4px;background:#f6f8fa;transition:border .2s linear}.news-page-load-more:hover{border:1px solid #1a7aff}.news-page-load-more__loading-failed{margin:20px 0 35px;text-align:center;font-size:15px;color:#d0021b}.news-page{color:#333}.news-page__container{display:-ms-flexbox;display:flex;-ms-flex-wrap:wrap;flex-wrap:wrap}.news-page__loading-failed{color:#d0021b}.news{margin:0 0 90px}.news__title{margin-bottom:5px;letter-spacing:-.1px;font-size:24px;line-height:30px;color:#333}.news__date{display:block;font-size:13px;opacity:.5}.news__content,.news__date{letter-spacing:0;line-height:25px;color:#333}.news__content{font-size:15px}.news__content h2{margin-top:30px;margin-bottom:0;letter-spacing:0;font-size:15px;font-weight:700;line-height:25px;color:#333}.news__content p{margin:25px 0}.news__content code{font-family:Courier New,serif}.news__content a{text-decoration:none;color:#1a7aff;transition:opacity .3s ease-out}.news__content a:hover{opacity:.5}.news__content mark{color:#ffce1e;background-color:transparent}.news__content img{display:block;max-width:100%}.news__share-label{margin-right:10px;letter-spacing:0;font-size:13px;line-height:25px;opacity:.5;color:#333}.news__footer{display:-ms-flexbox;display:flex;margin:28px 0 19px}.news__loading-failed{color:#d0021b}.ya-share2__item{transition:opacity .3s}.page-404{position:relative;width:100%;min-width:868px;height:calc(100vh - 70px);min-height:622px;padding-top:70px;background-color:#0077c4}.page-404__title{margin:108px 0 28px;text-align:center;font-size:48px;font-weight:300;color:#fff}.page-404__description{display:block;margin:28px 0;text-align:center;text-decoration:none;font-size:24px;font-weight:700;color:#fff}.page-404__moon{position:absolute;top:164px;left:21%}.page-404__cat{bottom:0;left:50%;width:319px;height:338px;transform:translateX(-50%)}.page-404__cat,.page-404__hills{position:absolute;background-repeat:no-repeat;background-position:50%}.page-404__hills{top:46%;width:100%;padding-top:14%;background-size:100%}:focus{outline:none}::-moz-focus-inner{border:0}.keynav :focus{box-shadow:0 0 0 2px #fff,0 0 0 4px #1a7aff}</style>
</head>
<body >
<div id="mount"><div class="app" data-reactroot="" data-reactid="1" data-react-checksum="1234929009"><!-- react-empty: 2 --><div data-reactid="3"><nav class="navbar" data-reactid="4"><div class="container navbar__links-wrapper" data-reactid="5"><a class="navbar__logo active" aria-current="true" href="/" data-reactid="6"><div class="navbar__arrow" style="background-image:url(//avatars.mds.yandex.net/get-bunker/50064/4d11a0d23580938980ea4ba1528affb6e3095f81/svg);" data-reactid="7"></div><div class="navbar__title" style="background-image:url(//avatars.mds.yandex.net/get-bunker/49769/499164f2c30cd977f20f5a2c4ffec254aa341298/svg);" data-reactid="8"></div></a><div class="navbar__links" data-reactid="9"><a class="navbar__link" href="/docs" target="_self" rel="noreferrer noopener" data-reactid="10">Documentation</a><a class="navbar__link" href="https://github.com/catboost" target="_blank" rel="noreferrer noopener" data-reactid="11">GitHub</a><a class="navbar__link" aria-current="false" href="/news" data-reactid="12">News</a><a class="navbar__link" aria-current="false" href="/#benchmark" data-reactid="13">Benchmarks</a><a class="navbar__link" href="https://forms.yandex.ru/surveys/10011699/?lang=en" target="_blank" rel="noreferrer noopener" data-reactid="14">Your Feedback</a><a class="navbar__link" aria-current="false" href="/#contacts" data-reactid="15">Contacts</a></div><ul class="navbar__social-links-container" data-reactid="16"><li class="navbar__social-link-li" data-reactid="17"><a target="_blank" rel="noopener noreferrer" href="https://t.me/catboost_en" class="navbar__social-link navbar__social-link_telegram" data-reactid="18"></a></li><li class="navbar__social-link-li" data-reactid="19"><a target="_blank" rel="noopener noreferrer" href="https://twitter.com/CatBoostML" class="navbar__social-link navbar__social-link_twitter" data-reactid="20"></a></li></ul></div></nav><div class="page-404" data-reactid="21"><h1 class="page-404__title" data-reactid="22">404 Not Found</h1><a class="page-404__description" href="/" data-reactid="23">← Return to home page</a><svg class="page-404__moon" width="86" height="86" data-reactid="24"><circle fill="#fff" cx="43" cy="43" r="43" data-reactid="25"></circle></svg><div class="page-404__hills" style="background-image:url(//avatars.mds.yandex.net/get-bunker/56833/85e015732dff89792151598170732b9982e64d4b/svg);" data-reactid="26"></div><div class="page-404__cat" style="background-image:url(//avatars.mds.yandex.net/get-bunker/50064/cae01a5f7083c686dec7d67c39a83e613a0b002d/svg);" data-reactid="27"></div></div><div data-reactid="28"></div></div><!-- react-empty: 29 --></div></div>
<script src="//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/vendor.js"></script>
<script src="//yastatic.net/s3/distribution/promo/promo-catboost-2017/v0.3.2/build/index.build.js"></script>
<script nonce="DqdhRCAqFwIwcRU8FqSbig==">__init__({"api":{"captcha":"\/captcha\/","subscribe":"\/subscribe\/","news":"\/api\/news\/"},"news":{"items":[],"current":{}}}, {"bunker":{"":{"schemas":[{"href":"\/promo-catboost-2017\/.schemas\/features","description":"feature"},{"href":"\/promo-catboost-2017\/.schemas\/about","description":"section \"About\""},{"href":"\/promo-catboost-2017\/.schemas\/footer\/link","description":"links in footer"},{"href":"\/promo-catboost-2017\/.schemas\/title","description":"Title"},{"href":"\/promo-catboost-2017\/.schemas\/news","description":"News"},{"href":"\/promo-catboost-2017\/.schemas\/navbar","description":"NavLinks"},{"href":"\/promo-catboost-2017\/.schemas\/meta","description":"meta"},{"href":"\/promo-catboost-2017\/.schemas\/head","description":"head"},{"href":"\/promo-catboost-2017\/.schemas\/benchmark","description":"benckmark"},{"href":"\/promo-catboost-2017\/.schemas\/og","description":"OpenGraph"},{"href":"\/promo-catboost-2017\/.schemas\/social-links","description":"Social links"},{"href":"\/promo-catboost-2017\/.schemas\/redirections","description":"Redirection Paths"},{"href":"\/promo-catboost-2017\/.schemas\/newsletter","description":"Newsletter"},{"href":"\/promo-catboost-2017\/.schemas\/errors","description":"News Errors Texts"},{"href":"\/promo-catboost-2017\/.schemas\/news\/texts","description":"News Texts"},{"href":"\/promo-catboost-2017\/.schemas\/benchmark\/quality","description":"Quality Benchmarks"},{"href":"\/promo-catboost-2017\/.schemas\/benchmark\/learningspeed","description":"Learning Speed Benchmarks"},{"href":"\/promo-catboost-2017\/.schemas\/sender","description":"Sender Configs"},{"href":"\/promo-catboost-2017\/.schemas\/pagexxx","description":"Page XXX"},{"href":"\/promo-catboost-2017\/.schemas\/contacts","description":"Contacts"},{"href":"\/promo-catboost-2017\/.schemas\/navbar\/social","description":"Social links in navbar"}]},"about":{"title":"About","text":"CatBoost is an algorithm for \<b\>gradient boosting on decision trees\<\/b\>. It is developed by Yandex researchers and engineers, and is used for search, recommendation systems, personal assistant, self-driving cars, weather prediction and many other tasks at Yandex and in other companies, including CERN, Cloudflare, Careem taxi. It is in open-source and can be used by anyone.","video":"https:\/\/www.youtube.com\/embed\/s8Q_orF4tcI?rel=0&showinfo=0"},"benchmark":{"data":{"dataColumns":["CatBoost","LightGBM","XGBoost","H2O"],"legendTexts":{"good":"Better","bad":"Worse","subtextRegression":"Figures in this table represent R2 values.","subtextClassification":"Figures in this table represent Logloss values (lower is better) for Classification mode.","aboutPercentage":"Percentage is metric difference measured against tuned CatBoost results."},"rows":[{"classification":{"title":"Adult","link":"https:\/\/archive.ics.uci.edu\/ml\/datasets\/Adult","data":["0.26974","0.27298","0.27602","0.28716","0.27542","0.28009","0.27510","0.27607"]},"regression":{"title":"Liberty property","link":"https:\/\/www.kaggle.com\/c\/liberty-mutual-group-property-inspection-prediction","data":["0.102043 ","0.100615 ","0.097696 ","-0.030386 ","0.083583 ","0.065600 ","0.09930","0.06887"]}},{"classification":{"title":"Amazon","link":"https:\/\/www.kaggle.com\/c\/amazon-employee-access-challenge","data":["0.13772","0.13811","0.16360","0.16716","0.16327","0.16536","0.16264","0.16950"]},"regression":{"title":"Liberty Fire","link":"https:\/\/www.kaggle.com\/c\/liberty-mutual-fire-peril","data":["0.000328","0.000355","0.000319","-0.000448","0.000178","-0.008828","0","-0.000023"]}},{"classification":{"title":"Click prediction","link":"http:\/\/www.kdd.org\/kdd-cup\/view\/kdd-cup-2012-track-2","data":["0.39090","0.39112","0.39633","0.39749","0.39624","0.39764","0.39759","0.39785"]},"regression":{"title":"","link":"","data":[]}},{"classification":{"title":"KDD appetency","link":"http:\/\/www.kdd.org\/kdd-cup\/view\/kdd-cup-2009\/Data","data":["0.07151","0.07138","0.07179","0.07482","0.07176","0.07466","0.07246","0.07355"]},"regression":{"title":"","link":"","data":[]}},{"classification":{"title":"KDD churn","link":"http:\/\/www.kdd.org\/kdd-cup\/view\/kdd-cup-2009\/Data","data":["0.23129","0.23193","0.23205","0.23565","0.23312","0.23369","0.23275","0.23287"]},"regression":{"title":"","link":"","data":[]}},{"classification":{"title":"KDD internet","link":"https:\/\/kdd.ics.uci.edu\/databases\/internet_usage\/internet_usage.html","data":["0.20875","0.22021","0.22315","0.23627","0.22532","0.23468","0.22209","0.24023"]},"regression":{"title":"","link":"","data":[]}},{"classification":{"title":"KDD upselling","link":"http:\/\/www.kdd.org\/kdd-cup\/view\/kdd-cup-2009\/Data","data":["0.16613","0.16674","0.16682","0.17107","0.16632","0.16873","0.16824","0.16981"]},"regression":{"title":"","link":"","data":[]}},{"classification":{"title":"KDD 98","link":"https:\/\/kdd.ics.uci.edu\/databases\/kddcup98\/kddcup98.html","data":["0.19467","0.19479","0.19576","0.19837","0.19568","0.19795","0.19539","0.19607"]},"regression":{"title":"","link":"","data":[]}},{"classification":{"title":"Kick prediction","link":"https:\/\/www.kaggle.com\/c\/DontGetKicked","data":["0.28479","0.28491","0.29566","0.29877","0.29465","0.29816","0.29481","0.29635"]},"regression":{"title":"","link":"","data":[]}}]},"quality":{"title":"Quality","tunedText":"Tuned","defaultText":"Default","data":{"columns":["CatBoost","LightGBM","XGBoost","H2O"],"rows":[{"name":"Adult","link":"https:\/\/archive.ics.uci.edu\/ml\/datasets\/Adult","data":{"tuned":["0.26974","0.27602","0.27542","0.27510"],"default":["0.27298","0.28716","0.28009","0.27607"]}},{"name":"Amazon","link":"https:\/\/www.kaggle.com\/c\/amazon-employee-access-challenge","data":{"tuned":["0.13772","0.16360","0.16327","0.16264"],"default":["0.13811","0.16716","0.16536","0.16950"]}},{"name":"Click prediction","link":"http:\/\/www.kdd.org\/kdd-cup\/view\/kdd-cup-2012-track-2","data":{"tuned":["0.39090","0.39633","0.39624","0.39759"],"default":["0.39112","0.39749","0.39764","0.39785"]}},{"name":"KDD appetency","link":"http:\/\/www.kdd.org\/kdd-cup\/view\/kdd-cup-2009\/Data","data":{"tuned":["0.07151","0.07179","0.07176","0.07246"],"default":["0.07138","0.07482","0.07466","0.07355"]}},{"name":"KDD churn","link":"http:\/\/www.kdd.org\/kdd-cup\/view\/kdd-cup-2009\/Data","data":{"tuned":["0.23129","0.23205","0.23312","0.23275"],"default":["0.23193","0.23565","0.23369","0.23287"]}},{"name":"KDD internet","link":"https:\/\/kdd.ics.uci.edu\/databases\/internet_usage\/internet_usage.html","data":{"tuned":["0.20875","0.22315","0.22532","0.22209"],"default":["0.22021","0.23627","0.23468","0.24023"]}},{"name":"KDD upselling","link":"http:\/\/www.kdd.org\/kdd-cup\/view\/kdd-cup-2009\/Data","data":{"tuned":["0.16613","0.16682","0.16632","0.16824"],"default":["0.16674","0.17107","0.16873","0.16981"]}},{"name":"KDD 98","link":"https:\/\/kdd.ics.uci.edu\/databases\/kddcup98\/kddcup98.html","data":{"tuned":["0.19467","0.19576","0.19568","0.19539"],"default":["0.19479","0.19837","0.19795","0.19607"]}},{"name":"Kick prediction","link":"https:\/\/www.kaggle.com\/c\/DontGetKicked","data":{"tuned":["0.28479","0.29566","0.29465","0.29481"],"default":["0.28491","0.29877","0.29816","0.29635"]}}],"legend":["Figures in this table represent Logloss values (lower is better) for Classification mode.","Percentage is metric difference measured against tuned CatBoost results."]}},"speed":{"title":"Learning speed","largeDatasetText":"Epsilon dataset","smallDatasetText":"Higgs dataset","units":{"sec":"sec"},"data":{"columns":["CatBoost","XGBoost","LightGBM"],"rows":[{"name":"CPU (Xeon E5-2660v4)","link":"","data":{"large":[527,4339,1146],"small":[770,881,438]}},{"name":"GTX 1080Ti (11GB)","link":"","data":{"large":[18,890,110],"small":[32,91,94]}}],"legend":{"small":["Dataset Higgs (4M samples, 28 features). Parameters: 128 bins, 64 leafs, 400 iterations."],"large":["Dataset Epsilon (400K samples, 2000 features). Parameters: 128 bins, 64 leafs, 400 iterations."]}}},"title":"Benchmarks"},"contacts":{"title":"Contacts","contactsLinks":["Report an issue with CatBoost on \<a href='https:\/\/github.com\/catboost\/catboost\/issues' target='_blank'\>GitHub\<\/a\>.","Ask a question on \<a href='https:\/\/stackoverflow.com\/questions\/tagged\/catboost' target='_blank'\>Stack Overflow\<\/a\> with the catboost tag, we monitor this for new questions.","Join Telegram chat to discuss with real users in \<a href='https:\/\/t.me\/catboost_en' target='_blank'\>English\<\/a\> or in \<a href='https:\/\/t.me\/catboost_ru' target='_blank'\>Russian\<\/a\>."]},"features":[{"text":"Reduce time spent on parameter tuning, because CatBoost provides great results with default parameters","title":"Great quality without parameter tuning","color":"#ffcd1f","imageID":5},{"text":"Improve your training results with CatBoost that allows you to use non-numeric factors, instead of having to pre-process your data or spend time and effort turning it to numbers. ","title":"Categorical features support","color":"#1a7aff","imageID":1},{"text":"Train your model on a fast implementation of gradient-boosting algorithm for GPU. Use a multi-card configuration for large datasets.","title":"Fast and scalable GPU version","color":"#008000","imageID":2},{"text":"Reduce overfitting when constructing your models with a novel gradient-boosting scheme.","title":"Improved accuracy","color":"#8b00ff","imageID":3},{"text":"Apply your trained model quickly and efficiently even to latency-critical tasks using CatBoost's model applier","title":"Fast prediction","color":"#ff0000","imageID":4}],"head":{"buttons":[{"text":"How to install","url":"\/docs\/concepts\/python-installation.html","target":"_blank","goal":"clck_install"},{"text":"Tutorials","url":"\/docs\/concepts\/tutorials.html","target":"_blank","goal":"clck_tutorials"}],"text":"CatBoost is a high-performance open source library for gradient boosting on decision trees"},"images":{"404":{"cat":"\/\/avatars.mds.yandex.net\/get-bunker\/50064\/cae01a5f7083c686dec7d67c39a83e613a0b002d\/svg","hills":"\/\/avatars.mds.yandex.net\/get-bunker\/56833\/85e015732dff89792151598170732b9982e64d4b\/svg"},"about":{"title":"About","video-overlay":"\/\/avatars.mds.yandex.net\/get-bunker\/49769\/f69db194d431ffdbdd913eb527606985eae5cf7e\/orig"},"captcha-refresh":"\/\/avatars.mds.yandex.net\/get-bunker\/56833\/a648f12fd3e5e70a68aafed1d2b026d4ad803b72\/svg","captcha-voice":"\/\/avatars.mds.yandex.net\/get-bunker\/118781\/7f177f3cf2404fad5d71a94d421bf9a942b56bb5\/svg","cat":{"image":"\/\/avatars.mds.yandex.net\/get-bunker\/50064\/e22377275c7ecb711d82156f20bfd29c589aec60\/orig"},"check":"\/\/avatars.mds.yandex.net\/get-bunker\/118781\/a8a099651ffa021037b0fb87a85677798fb56638\/svg","favicons":{"favicon.png":"\/\/avatars.mds.yandex.net\/get-bunker\/49769\/c4b06f2b169c7596da4143d5f951d925b17c413e\/orig","favicon-120.png":"\/\/avatars.mds.yandex.net\/get-bunker\/49769\/96a4132b5fccfbe6a8a18a71279626bcabe2ba62\/orig","favicon-60.png":"\/\/avatars.mds.yandex.net\/get-bunker\/49769\/a74e0923e4e58febf68787226ef24769aea5a2e5\/orig"},"features":{"1":"\/\/avatars.mds.yandex.net\/get-bunker\/49769\/fabea970320f9e06b9732d9cfc513fa5e5439207\/orig","2":"\/\/avatars.mds.yandex.net\/get-bunker\/118781\/050858bb850a3f19f3afdf4023cf555ddcee9550\/orig","3":"\/\/avatars.mds.yandex.net\/get-bunker\/998550\/dd5472447e03c29a5de08b73533ccdeca6baa240\/orig","4":"\/\/avatars.mds.yandex.net\/get-bunker\/998550\/22acd52145d214cdf98e0f2ccfbdf3e9cd16e049\/orig","5":"\/\/avatars.mds.yandex.net\/get-bunker\/120922\/06a8bf80af45438dba510e8d027ae22dd7d9dd7b\/orig"},"footer":{"share":"\/\/avatars.mds.yandex.net\/get-bunker\/56833\/dba868860690e7fe8b68223bb3b749ed8a36fbce\/orig"},"head":"\/\/avatars.mds.yandex.net\/get-bunker\/128809\/2c9167204a2a0432c56806fac0532f7a0c1d9f3f\/orig","navbar":{"logo":"\/\/avatars.mds.yandex.net\/get-bunker\/50064\/4d11a0d23580938980ea4ba1528affb6e3095f81\/svg","title":"\/\/avatars.mds.yandex.net\/get-bunker\/49769\/499164f2c30cd977f20f5a2c4ffec254aa341298\/svg"},"news":{"auc_iter":"\/\/avatars.mds.yandex.net\/get-bunker\/135516\/72f3d00ee182eb01d5ce0155378e710482540331\/orig","auc_time":"\/\/avatars.mds.yandex.net\/get-bunker\/118781\/1af14d79f7a0257a5a47bbb6b76e71be5108b271\/orig","collider":"\/\/avatars.mds.yandex.net\/get-bunker\/118781\/05a88813e6cf89c138373b897ef2581fc2c33be9\/orig","collider-560x315":"\/\/avatars.mds.yandex.net\/get-bunker\/61205\/c83cdcae23382225cba7e602a21be5989c467987\/orig","fast_inf":"\/\/avatars.mds.yandex.net\/get-bunker\/120922\/58e11eb206a498bbea44041cb3d7e5e2e181c6ae\/orig","gpu_gtc18":"\/\/avatars.mds.yandex.net\/get-bunker\/118781\/7592f70cef5e12a005a7fd5a67fd34590293cc15\/orig","icml":"\/\/avatars.mds.yandex.net\/get-bunker\/50064\/5f50486ae148954da2e6c5b795820c5ec3701a6d\/orig","icml-560x315":"\/\/avatars.mds.yandex.net\/get-bunker\/135516\/e0711c75099be91e2360c3acb8be916fd29f8d43\/orig","shap_values":"\/\/avatars.mds.yandex.net\/get-bunker\/49769\/47b1e47839734be08e9cb86deb3375999888a593\/orig","tensorboard":"\/\/avatars.mds.yandex.net\/get-bunker\/49769\/0a1456479b7d8a1be8056ffa05bfcebb98387f9e\/orig","tensorboard-560x315":"\/\/avatars.mds.yandex.net\/get-bunker\/49769\/a22af2c5c6afb71ab610701020ff1651b9b99c30\/orig"}},"meta":{"title":"CatBoost - open-source gradient boosting library","description":"CatBoost is an open-source gradient boosting on decision trees library with categorical features support out of the box, successor of the MatrixNet algorithm developed by Yandex.","keywords":"catboost, matrixnet, boosting, gradient boosting, gradient boosting on decision trees, decision trees, classification, regression, LightGBM, XGBoost","image":""},"navbar":[{"id":1,"text":"Documentation","url":"\/docs","ym":"clck_doc","isInternal":false,"target":"_self"},{"id":2,"text":"GitHub","url":"https:\/\/github.com\/catboost","ym":"clck_github","isInternal":false,"target":"_blank"},{"id":3,"text":"News","url":"\/news","ym":"clck_news","isInternal":true,"target":"_self"},{"id":4,"text":"Benchmarks","url":"\/#benchmark","ym":"clck_benchmark","isInternal":true,"target":"_self"},{"id":5,"text":"Your Feedback","url":"https:\/\/forms.yandex.ru\/surveys\/10011699\/?lang=en","ym":"clck_survey","isInternal":false,"target":"_blank"},{"id":6,"text":"Contacts","url":"\/#contacts","ym":"clck_contacts","isInternal":true,"target":"_self"}],"news":{"errors":{"loadMoreErrorText":"Failed to get news, please click the button to try again","articleErrorText":"Failed to get the article, please refresh the page to try again.","lastNewsErrorText":"Failed to get news, please refresh the page to try again."},"list":[{"title":"New ways to explore your data","date":"Apr 20, 2018","preview":"New superb tool for exploring feature importance, new algorithm for finding most influential training samples, possibility to save your model as cpp or python code and more. Check CatBoost v0.8 details inside!","text":"It’s time to release CatBoost v0.8. The aim of this release - efficient tools for data and model exploration.\n\<br\>\<br\>\nFirst of all, catboost now calculates per object feature importances using SHAP values algorithm from the ‘Consistent feature attribution for tree ensembles’ \<a href='https:\/\/arxiv.org\/pdf\/1706.06060.pdf' target='_blank'\>paper\<\/a\>. As you can see on the picture below it's very easy to understand what is the influence of each feature on a given object. See \<a href='https:\/\/github.com\/catboost\/catboost\/blob\/master\/catboost\/tutorials\/advanced_tutorials\/shap_values_tutorial.ipynb' target='_blank'\>tutorial\<\/a\> for more details.\n\<img alt=\"SHAP values\" bunkerImageKey=\"news.shap_values\"\>\nSecondly, CatBoost now has a new algorithm for finding most influential training samples for a given object. This mode calculates the effect of objects from the train dataset on the optimized metric values for the objects from the input dataset:\n\<br\>- Positive values reflect that the optimized metric increases.\n\<br\>- Negative values reflect that the optimized metric decreases.\n\<br\>\nThe higher the deviation from 0, the bigger the impact that an object has on the optimized metric. The method is an implementation of the approach described in the 'Finding Influential Training Samples for Gradient Boosted Decision Trees' \<a href='https:\/\/arxiv.org\/pdf\/1802.06640.pdf' target='_blank'\>paper\<\/a\>. See \<a href='https:\/\/tech.yandex.com\/catboost\/doc\/dg\/concepts\/python-reference_catboost_get_object_importance-docpage\/' target='_blank'\>get_object_importance\<\/a\> model method in Python package and \<a href='https:\/\/tech.yandex.com\/catboost\/doc\/dg\/concepts\/cli-reference_ostr-calc-docpage\/' target='_blank'\>ostr\<\/a\> mode in cli-version. Tutorial for Python is also \<a href='https:\/\/github.com\/catboost\/catboost\/blob\/master\/catboost\/tutorials\/advanced_tutorials\/catboost_object_importance_tutorial.ipynb' target='_blank'\>available\<\/a\>.\n\<br\>\<br\>\nThird cool staff in 0.8 release is ’save model as code’ feature. For now you could save model as Python code with categorical features and as C++ code without categorical features (сategorical features support for C++ is coming soon). Use \<i\>--model-format CPP,Python\<\/i\> in cli-version and \<i\>model.save_model(OUTPUT_PYTHON_MODEL_PATH, format=\"python\")\<\/i\> in Python.\n\<br\>\<br\>\nTo find out more details check out release notices on \<a href='https:\/\/github.com\/catboost\/catboost\/releases' target='_blank'\>GitHub\<\/a\>.\n\<br\>\<br\>\nAs usual we are eager to see your \<a href='https:\/\/github.com\/catboost\/catboost\/issues' target='_blank'\>feedback\<\/a\> and contribution.","id":"version_0_8","order":0},{"title":"CatBoost on GPU talk at GTC 2018","date":"Mar 26, 2018","preview":"Come and listen our talk about the fastest implementation of Gradient Boosting for GPU at the GTC 2018 Silicon Valley! GTC will take place on March 26-29 and will provide an excellent opportunity to get more details about CatBoost performance on GPU.","text":"Vasily Ershov CatBoost lead developer, \<a href=\"https:\/\/2018gputechconf.smarteventscloud.com\/connect\/sessionDetail.ww?SESSION_ID=152312\" target=\"_blank\"\>will talk\<\/a\> on GTC 2018 about fastest gradient boosting implementation on GPU.\n\<br\>\<br\>\nHe'll provide a brief overview of problems which could be solved with CatBoost, discuss challenges and key optimizations in the most significant computation blocks and describe how one can efficiently build histograms in shared memory to construct decision trees and how to avoid atomic operation during this step. Also he'll provide benchmarks that shows that our GPU implementation is five to 40 times faster compared to CPU. And finally he'll talk about performance comparison against GPU implementations of gradient boosting in other open-source libraries.\n\<br\>\<br\>\n\<img alt=\"GPU comparison\" bunkerImageKey=\"news.gpu_gtc18\"\>\n\<br\>\nPicture contains learning speed on GPU comparison between CatBoost, XGBoost and LightGBM on epsilon dataset. Notice XGBoost GPU implementation doesn’t support V100 cards.\n\<br\>\<br\>\nTalk scheduled for \<b\>March 27, 1:00 PM - Room 231\<\/b\>. Don’t miss the chance to listen about best-in-class gradient boosting implementation on GPU and ask any question.","id":"gtc2018","order":0},{"title":"Best in class inference and a ton of speedups","date":"Jan 31, 2018","preview":"New version of CatBoost has industry fastest inference implementation. It's 35 times faster than open-source alternatives and completely production ready. Furthermore 0.6 release contains a lot of speedups and improvements. Find more inside.","text":"CatBoost version 0.6 has a lot speedups and improvements. Most valuable improvement at the moment is the release of industry fastest inference implementation.\n\<br\>\<br\>\n\<img alt=\"FastInference\" bunkerImageKey=\"news.fast_inf\"\>\n\n\<p\>\<b\>Fast inference\<\/b\>\<\/p\>\nCatBoost uses oblivious trees as base predictors. In oblivious trees each leaf index can be encoded as a binary vector with length equal to the depth of the tree. This fact is widely used in CatBoost model evaluator: we first binarize all used float features, statistics and one-hot encoded features and then use binary features to calculate model predictions. That vectors can be built in a data parallel manner with SSE intrinsics. This results in a much faster applier than all existing ones as shown in our comparison below.\n\n\<p\>\<b\>CatBoost applier vs LightGBM vs XGBoost\<\/b\>\<\/p\>\n\nWe used LightGBM, XGBoost and CatBoost models for \<a href=\"https:\/\/www.csie.ntu.edu.tw\/~cjlin\/libsvmtools\/datasets\/binary.html\" target=\"blank\"\>Epsilon\<\/a\> (400K samples, 2000 features) dataset trained as described in our previous \<a href=\"https:\/\/catboost.yandex\/news#version_0_3\" target=\"blank\"\>benchmarks\<\/a\>. For each model we limit number of trees used for evaluation to 8000 to make results comparable for the reasons described above. Thus this comparison gives only some insights of how fast the models can be applied. For each algorithm we loaded test dataset in Python, converted it to the algorithm internal representation and measured wall-time of model predictions on Intel Xeon E5-2660 CPU with 128GB RAM. The results are presented in the table below.\n\<br\>\<br\>\n\<table width=\"100%\" border=\"1\"\>\n \<tbody\>\<tr align=\"left\"\>\<th\>\<\/th\>\<th\>1 thread\<\/th\>\<th\>32 thread\<\/th\>\<\/tr\>\n \<tr\>\<td\>XGBoost\<\/td\>\<td\>71 sec (x39)\<\/td\>\<td\>4,5 sec (x31)\<\/td\>\<\/tr\>\n \<tr\>\<td\>LightGBM\<\/td\>\<td\>88 sec (x48)\<\/td\>\<td\>17,1 sec(x118)\<\/td\>\<\/tr\>\n \<tr\>\<td\>CatBoost\<\/td\>\<td\>\<b\>1,83 sec\<\/b\>\<\/td\>\<td\>\<b\>0,145 sec\<\/b\>\<\/td\>\<\/tr\>\n\<\/tbody\>\<\/table\>\n\<br\>\nFrom this we can see that on similar sizes of ensembles CatBoost can be applied about 35 and 83 times faster than XGBoost and LightGBM respectively.\n\n\<p\>\<b\>Speedups\<\/b\>\<\/p\>\nCatBoost team spent a lot of effort to speedup different parts of library. For now the list is below:\n\<ul\>\n \<li\>43% speedup for training on large datasets.\<\/li\>\n \<li\>15% speedup for \<i\>QueryRMSE\<\/i\> and calculation of querywise metrics.\<\/li\>\n \<li\>Large speedups when using binary categorical features.\<\/li\>\n \<li\>Significant (x200 on 5k trees and 50k lines dataset) speedup for plot and stage predict calculations in cmdline.\<\/li\>\n \<li\>Compilation time speedup.\<\/li\>\n\<\/ul\>\nPlease take notice, we added many synonyms to our parameter names, now it is more convenient to try CatBoost if you are used to some other library.\n\<br\>\<br\>\nOther improvements, bug fixes as well as builds you could find in \<a href=\"https:\/\/github.com\/catboost\/catboost\/releases\/tag\/v0.6\" target=\"blank\"\>release\<\/a\> on GitHub.\n\<br\>\<br\>\nFeel free to drop us \<a href=\"https:\/\/github.com\/catboost\/catboost\/issues\" target=\"blank\"\>issue\<\/a\> or contribute to the project.","id":"version_0_6","order":0},{"title":"Extremely fast learning on GPU has arrived!","date":"Nov 2, 2017","preview":"CatBoost version 0.3 brings efficient support of distributed training on GPU! One server with 8 GPUs can process as much data as few hundreds of CPU servers and will work much faster. Even with a single GPU you will get up to 40x speed up of your training. Check out our benchmarks inside and download new version on GitHub.","text":"\<p\>We're excited to announce the CatBoost 0.3 release with GPU support. It's incredibly fast!\<\/p\>\n\<ul\>\n \<li\>Training on a single GPU outperforms CPU training up to 40x times on large datasets.\<\/li\>\n \<li\>CatBoost efficiently supports multi-card per unit configuration. A configuration with a single server with 8 GPUs outperforms a configuration with hundreds of CPUs by execution time.\<\/li\>\n \<li\>We compared our GPU implementation with competitors. And it is 2 times faster then LightGBM and more then 20 times faster then XGBoost.\<\/li\>\n \<li\>Finally CatBoost GPU Python wrapper is very easy to use.\<\/li\>\n\<\/ul\>\n\<p\>To prove our words we prepared a set of benchmarks below. There we compared:\<\/p\>\n\<ul\>\n \<li\>CPU vs GPU training speed of CatBoost\<\/li\>\n \<li\>GPU training performance of CatBoost, XGBoost and LightGBM\<\/li\>\n\<\/ul\>\n\<p\>\<b\>CatBoost CPU vs. GPU\<\/b\>\<\/p\>\n\<p\>\<i\>Configuration:\<\/i\> dual-socket server with 2 Intel Xeon CPU (E5-2650v2, 2.60GHz) and 256GB RAM\<\/p\>\n\<p\>\<i\>Methodology:\<\/i\> CatBoost was started in 32 threads (equal to number of logical cores). GPU implementation was run on several servers with different GPU types. Our GPU implementation doesn’t require multi-core server for high performance, so different CPU and machines should not significantly affect GPU benchmark results.\n\<\/p\>\n\<p\>\<i\>Dataset #1:\<\/i\> \<a href=\"https:\/\/www.kaggle.com\/c\/criteo-display-ad-challenge\/data\" target=\"blank\"\>Criteo\<\/a\> (36M samples, 26 categorical, 13 numerical features) to benchmark our categorical features support.\<\/p\>\n\<table width=\"100%\" border=\"1\"\>\n \<tbody\>\<tr align=\"left\"\>\<th\>Type\<\/th\>\<th\>128 bins (sec)\<\/th\>\<\/tr\>\n \<tr\>\<td\>CPU\<\/td\>\<td\>1060\<\/td\>\<\/tr\>\n \<tr\>\<td\>K40\<\/td\>\<td\>373\<\/td\>\<\/tr\>\n \<tr\>\<td\>GTX 1080Ti (11GB) \<\/td\>\<td\>301\<\/td\>\<\/tr\>\n \<tr\>\<td\>2xGTX 1080 (8GB)\<\/td\>\<td\>285\<\/td\>\<\/tr\>\n \<tr\>\<td\>P40\<\/td\>\<td\>123\<\/td\>\<\/tr\>\n \<tr\>\<td\>P100-PCI\<\/td\>\<td\>82\<\/td\>\<\/tr\>\n \<tr\>\<td\>V100-PCI\<\/td\>\<td\>69.8\<\/td\>\<\/tr\>\n \<\/tbody\>\<\/table\>\n\<p\>As you can see CatBoost GPU version significantly outperforms CPU training time even on old generation GPU (Tesla K40) and gains impressive x15 speed up on flagship NVIDIA V100 card.\<\/p\>\n\<p\>\<i\>Dataset #2:\<\/i\> \<a href= \"http:\/\/www.csie.ntu.edu.tw\/~cjlin\/libsvmtools\/datasets\/binary.html\" target=\"blank\"\>Epsilon\<\/a\> (400K samples, 2000 features) to benchmark our performance on dense numerical dataset. In the table we report training time for different levels of binarizations: default 128 bins and 32 bins which is often sufficient. Note that Epsilon dataset has not enough samples to fully utilize GPU, and with bigger datasets we observe up to 40x speed ups.\<\/p\>\n\<table width=\"100%\" border=\"1\"\>\n \<tbody\>\<tr align=\"left\"\>\<th\>Type\<\/th\>\<th\>128 bins (sec)\<\/th\>\<th\>32 bins (sec)\<\/th\>\<\/tr\>\n \<tr\>\<td\>CPU\<\/td\>\<td\>713\<\/td\>\<td\>653\<\/td\>\<\/tr\>\n \<tr\>\<td\>K40\<\/td\>\<td\>547\<\/td\>\<td\>248\<\/td\>\<\/tr\>\n \<tr\>\<td\>GTX 1080 (8GB)\<\/td\>\<td\>194\<\/td\>\<td\>120\<\/td\>\<\/tr\>\n \<tr\>\<td\>P40\<\/td\>\<td\>162\<\/td\>\<td\>91\<\/td\>\<\/tr\>\n \<tr\>\<td\>GTX 1080Ti (11GB)\<\/td\>\<td\>145\<\/td\>\<td\>88\<\/td\>\<\/tr\>\n \<tr\>\<td\>P100-PCI\<\/td\>\<td\>127\<\/td\>\<td\>70\<\/td\>\<\/tr\>\n \<tr\>\<td\>V100-PCI\<\/td\>\<td\>77\<\/td\>\<td\>49\<\/td\>\<\/tr\>\n \<\/tbody\>\<\/table\>\n\<p\>\<b\>GPU training performance: comparison with baselines\<\/b\>\<\/p\>\n\<p\>\<i\>Configuration:\<\/i\> NVIDIA P100 accelerator, dual-core Intel Xeon E5-2660 CPU and 128GB RAM\<\/p\>\n\<p\>\<i\>Dataset:\<\/i\> Epsilon (400K samples for train, 100K samples for test).\<\/p\>\n\<p\>\<i\>Libraries:\<\/i\> CatBoost, LightGBM, XGBoost (we use histogram-based version, exact version is very slow)\<\/p\>\n\<p\>\<i\>Methodology:\<\/i\> We measured mean tree construction time one can achieve without using feature subsampling and\/or bagging. For XGBoost and CatBoost we use default tree depth equal to 6, for LightGBM we set leafs count to 64 to have more comparable results. We set bin to 15 for all 3 methods. Such bin count gives the best performance and the lowest memory usage for LightGBM and CatBoost (128-255 bin count usually leads both algorithms to run 2-4 times slower). For XGBoost we could use even smaller bin count but performance gains compared to 15 bins are too small to account for. All algorithms were run with 16 threads, which is equal to hardware core count. \<\/p\>\n\<p\>By default CatBoost uses \<a href=\"https:\/\/tech.yandex.com\/catboost\/doc\/dg\/concepts\/algorithm-main-stages_fighting-biases-docpage\/\" target=\"blank\"\>bias-fighting scheme\<\/a\> . This scheme is by design 2-3 times slower then classical boosting approach. CatBoost GPU implementation contains a mode based on classic scheme for those who need best training performance. We used classic scheme mode in our benchmark.\<\/p\>\n\<img alt=\"AUCvsNumber\" width=\"100%\" bunkerImageKey=\"news.auc_iter\"\>\n\<p\>Figure 1. AUC vs Number of trees\<\/p\>\n\<img alt=\"AUCvsTime\" width=\"100%\" bunkerImageKey=\"news.auc_time\"\>\n\<p\>Figure 2. AUC vs Time\<\/p\>\n\<p\>We set such learning rate that algorithms start to overfit approximately after 8000 rounds (learning curves are displayed at figure above, quality of obtained models differs by approximately 0.5%). We measured time to train ensembles of 8000 trees. Mean tree construction time for CatBoost was 17.9ms, for XGBoost 488ms, for LightGBM 40ms. As you can see CatBoost 2 times faster then LightGBM and 20 times faster then XGBoost.\<\/p\>\n\<p\>Don’t forget to examine CatBoost GPU \<a href=\"https:\/\/tech.yandex.com\/catboost\/doc\/dg\/features\/training-on-gpu-docpage\/#training-on-gpu\" target=\"blank\"\>documentation\<\/a\>. As usual, you could find all the code on \<a href=\"https:\/\/github.com\/catboost\" target=«blank\"\>GitHub\<\/a\>.\<\/p\>\n\<p\>Any contribution and \<a href=\"https:\/\/github.com\/catboost\/catboost\/issues\" target=\"blank\"\>issues\<\/a\> are appreciated!\<\/p\>","id":"version_0_3","order":0},{"title":"Version 0.2 released","date":"Sep 14, 2017","preview":"We are proud to release CatBoost version 0.2. Speed, stability, quality and ton of other improvements are already published on GitHub. Find the full list of improvements below.","text":"\<p\>In last few weeks, the CatBoost team have implemented a bunch of improvements.\n\<ul\>\n\<li\>Training speedups: we have speed up the training by 20-30%.\<\/li\>\n\<li\>Accuracy improvement with categoricals: we have changed computation of statistics for categorical features, which leads to better quality.\<\/li\>\n\<li\>New type of overfitting detector: \<i\>Iter\<\/i\>. This type of detector was requested by our users. So now you can also stop training by a simple criterion: if after a fixed number of iterations there is no improvement of your evaluation function.\<\/li\>\n\<li\>TensorBoard support: this is another way of looking on the graphs of different error functions both during training and after training has finished. To look at the metrics you need to provide \<i\>train_dir\<\/i\> when training your model and then run \<i\>\"tensorboard --logdir={train_dir}\"\<\/i\>.\<\/li\>\n\<\/ul\>\n\<\/p\>\n\<p\>\n\<img alt=\"TensorBoard\" width=\"100%\" bunkerImageKey=\"news.tensorboard-560x315\"\>\n\<\/p\>\n\<p\>\n\<ul\>\n\<li\>Jupyter notebook improvements: for our Python library users that experiment with Jupyter notebooks, we have improved our visualisation tool. Now it is possible to save image of the graph. We also have changed scrolling behaviour so that it is more convenient to scroll the notebook.\<\/li\>\n\<li\>NaN features support: we also have added simple but effective way of dealing with NaN features. If you have some NaNs in the train set, they will be changed to a value that is less than the minimum value or greater than the maximum value in the dataset (this is configurable), so that it is guaranteed that they are in their own bin, and a split would separates NaN values from all other values. By default, no NaNs are allowed, so you need to use option \<i\>nan_mode\<\/i\> for that. When applying a model, NaNs will be treated in the same way for the features where NaN values were seen in train. It is not allowed to have NaN values in test if no NaNs in train for this feature were provided.\<\/li\>\n\<li\>Snapshotting: we have added snapshotting to our Python and R libraries. So if you think that something can happen with your training, for example machine can reboot, you can use \<i\>snapshot_file\<\/i\> parameter - this way after you restart your training it will start from the last completed iteration.\<\/li\>\n\<li\>R library changes: we have changed an R library interface and added \<a href=\"https:\/\/github.com\/catboost\/catboost\/blob\/master\/catboost\/tutorials\/catboost_r_tutorial.ipynb\" target=\"blank\"\>tutorial\<\/a\>.\<\/li\>\n\<li\>Logging customization: we have added \<i\>allow_writing_files\<\/i\> parameter. By default some files with logging and diagnostics are written on disc, but you can turn it off using by setting this flag to False.\<\/li\>\n\<li\>Multiclass mode improvements: we have added a new objective for multiclass mode - \<i\>MultiClassOneVsAll\<\/i\>. We also added \<i\>class_names\<\/i\> param - now you don't have to renumber your classes to be able to use multiclass. And we have added two new metrics for multiclass: \<i\>TotalF1\<\/i\> and \<i\>MCC\<\/i\> metrics.\nYou can use the metrics to look how its values are changing during training or to use overfitting detection or cutting the model by best value of a given metric.\<\/li\>\n\<li\>Cross-validation parameters changes: we changed overfitting detector parameters of CV in python so that it is same as those in training.\<\/li\>\n\<li\>CTR types: \<i\>MeanValue\<\/i\> =\> \<i\>BinarizedTargetMeanValue\<\/i\>.\<\/li\>\n\<li\>Any delimeters support: in addition to datasets in \<i\>tsv\<\/i\> format, CatBoost now supports files with any delimeters.\<\/li\>\n\<li\>New model format: CatBoost v0.2 model binary not compatible with previous versions.\<\/li\>\n\<\/ul\>\n\<\/p\>\n\<p\>\nWe also have improved stability of the library.\n\<\/p\>\n\<p\>Feel free to write us with \<a href=\"https:\/\/github.com\/catboost\/catboost\/issues\" target=\"blank\"\>issues\<\/a\> on GitHub and contribute to the project!\<\/p\>","id":"version_0_2","order":0},{"title":"CatBoost at ICML 2017","date":"July 20, 2017","preview":"Come and meet us at the 2017 ICML conference in Sydney! The 34th International Conference on Machine Learning will take place on August 6-11 and will provide an excellent opportunity to get a demo of CatBoost in action.","text":"\<p\>We will be happy to meet everyone at ICML in Sydney, Australia on August 6-11, 2017, where we will be showcasing CatBoost in the Yandex booth #16.\<\/p\>\n\<p\>Our team will be there to showcase the usage and applications of our new gradient-boosting machine learning library. We’ll be happy to demonstrate training CatBoost on a variety of datasets, and go through the tricks CatBoost uses to work well on categorical features. You will learn how to access the CatBoost library from the command line, or via API for Python, sklearn, R or caret, and how to monitor training in iPython Notebook using our visualization tool CatBoost Viewer. We will also let you in on the secret of how to score well in a Kaggle contest!\<\/p\>\n\<img width=\"100%\" alt=\"ICML\" bunkerImageKey=\"news.icml-560x315\"\>\n\<p\>We look forward to meeting you at our ICML stand in Sydney. Please drop by – we’ll even have some goodies to share!\<\/p\>","id":"ICML","order":0},{"title":"Large Hadron Collider particle identification","date":"July 18, 2017","preview":"CatBoost was used to improve the state-of-the-art performance of data processing system at LHCb, one of the experiments at the Large Hadron Collider. The data collected by the experiment is processed by CatBoost for individual collisions happening at rate of 40 million per second","text":"\<p\>The Large Hadron Collider beauty (LHCb) experiment is one of the four major experiments running at the Large Hadron Collider (LHC), the world’s largest and most powerful particle accelerator, operating at the European Organization for Nuclear Research (CERN). In order to perform high-level physics measurements, scientists need to analyse data from particle collisions recorded at a rate of 40 million times per second.\<\/p\>\n\<p\>These data represent “snapshots” of all the particles generated by collisions of LHC protons and flying through the volume of particle detectors placed around the proton-proton interaction region. In order to understand the entire picture of the underlying physics laws ruling the processes taking place in the collisions, it is extremely important to reconstruct the identity of each particle whose passage is recorded by the detectors. This is the main role of particle identification (PID) algorithms.\<\/p\>\n\n\<img alt=\"Collider\" bunkerImageKey=\"news.collider-560x315\"\>\n\n\<p\>Fast, reliable and accurate PID algorithms are crucial to selecting interesting data. In almost all 400 or so papers published by the LHCb collaboration, it is evident that these aspects of PID algorithms play a crucial role in important discoveries.\<\/p\>\n\n\<p\>To combine the information from the various subcomponents of the LHCb detector in an effort to achieve a more efficient PID performance, a team from the Yandex School of Data Analysis proposed the use of the new algorithm CatBoost. CatBoost is well suited to improve the accuracy of PID response because it works with different features types (including binary observables) and formats with state-of-the-art precision. The algorithm ideally meets LHCb requirements, working as a seamless complement with all inputs.\<\/p\>\n\n\<p\>The algorithm was trained using simulated collisions resembling those taking place at the LHCb proton-proton interaction point. The algorithm uses about 60 input features describing the geometrical position of the detected particles, the aggregated detector response and the kinematic properties of the detected tracks.\<\/p\>\n\n\<p\>After its implementation and deployment into LHCb codebase and event processing pipeline in June 2017, CatBoost’s best-in-class performance proved to improve accuracy without compromising efficiency. Initial tests show encouraging improvements in the identification of charged particles starting from the information that they release in the LHCb detector, with respect to other machine learning approaches available on the market. Ultimately, this new approach will lead to cleaner data to all particle physics experiments, making physicists’ work more efficient.\<\/p\>\n\n\<p\>After seeing these initial positive results, the LHCb team is planning further experimentation with CatBoost in other LHCb projects.\<\/p\>","id":"particle_identification","order":0},{"title":"CatBoost Now Available in Open Source","date":"July 18, 2017","preview":"The CatBoost source code is now available on GitHub under Apache License 2.0. In addition to the actual CatBoost algorithm, you can enjoy Python and R packages, as well as a comparison tool for popular gradient-boosting libraries.","text":"\<p\>Today, we are open-sourcing our gradient boosting library CatBoost. It is well-suited for training machine learning models on tasks where data is heterogeneous, i.e., is described by a variety of inputs, such as contents, historical statistics and outputs of other machine learning models. The new gradient-boosting algorithm is now available on \<a href=\"https:\/\/github.com\/catboost\/\"\>GitHub\<\/a\> under Apache License 2.0.\<\/p\>\n\<iframe width=\"560\" height=\"315\" src=\"https:\/\/www.youtube.com\/embed\/s8Q_orF4tcI\" frameborder=\"0\" allowfullscreen\>\<\/iframe\>\n\<p\>Developed by Yandex data scientists and engineers, it is the successor of the MatrixNet algorithm that is used within the company for a wide range of tasks, ranging from ranking search results and advertisements to weather forecasting, fraud detection, and recommendations. In contrast to MatrixNet, which uses only numeric data, CatBoost can work with non-numeric information, such as cloud types or state\/province. It can use this information directly, without requiring conversion of categorical features into numbers, which may yield better results compared with other gradient-boosting algorithms and also saves time. The range of CatBoost applications includes a variety of spheres and industries, from banking and weather forecasting, to recommendation systems and steel manufacturing.\<\/p\>\n\<p\>CatBoost supports Linux, Windows and macOS and can also be operated from a command line or via a user-friendly API for Python or R. In addition to open-sourcing our gradient-boosting algorithm, we are releasing our visualization tool CatBoost Viewer, which enables monitoring training processes in iPython Notebook or in a standalone mode. We are also equipping all CatBoost users with a tool for comparing results of popular gradient-boosting algorithms.\<\/p\>\n\<p\>“Yandex has a long history in machine learning. We have the best experts in the field. By open-sourcing CatBoost, we are hoping that our contribution into machine learning will be appreciated by the expert community, who will help us to advance its further development,” says Misha Bilenko, Head of Machine Intelligence and Research at Yandex.\<\/p\>\n\<p\>CatBoost has already been successfully tested in a variety of applications across a whole range of Yandex services, including weather forecasting for the Meteum technology, content ranking for the personal recommendations service Yandex Zen, and improving search results. Eventually, this algorithm will be rolled out to benefit the majority of Yandex services. Outside of Yandex, CatBoost is already being used by data scientists at the European Organization for Nuclear Research (CERN) to improve data processing performance in their Large Hadron Collider beauty experiment.\<\/p\>","id":"open_source","order":0}],"texts":{"fullNewsTitle":"Read the full version","loadMoreTitle":"More","goToNews":"Full news list","lastNewsTitle":"Latest News"},"title":"News"},"newsletter":{"enabled":false,"initial":{"title":"Newsletter","description":"Get updates about important news, case studies, and our latest solutions.","inputPlaceholder":"e-mail address","buttonText":"Subscribe","errorText":"Please enter a valid e-mail address","termsConditionsText":"Will be used in accordance with our \<a href=\"#\" target=\"_blank\"\>User Agreement\<\/a\>"},"captcha":{"title":"Enter the code and prove you are not a robot","refreshText":"Refresh code","captchaReaderText":"Vision Impaired","inputPlaceholder":"Type the code above","buttonText":"Confirm","errorText":"Please enter a valid code"},"successText":"Thank you\<\/br\>for your subscription","errorText":"Something went wrong. Please try again"},"og":{"title":"CatBoost - state-of-the-art open-source gradient boosting library with categorical features support","image":"https:\/\/avatars.mds.yandex.net\/get-bunker\/56833\/dba868860690e7fe8b68223bb3b749ed8a36fbce\/orig","description":"CatBoost - state-of-the-art open-source gradient boosting library with categorical features support, https:\/\/catboost.yandex\/ #catboost","description-twitter":"#CatBoost - state-of-the-art open-source gradient boosting library with categorical features support, "},"page404":{"title":"404 Not Found","description":"← Return to home page"},"redirections":[{"from":"\/news#open_source","to":"\/news\/catboost-now-available-in-open-source"},{"from":"\/news#particle_identification","to":"\/news\/large-hadron-collider-particle-identification"},{"from":"\/news#ICML","to":"\/news\/catboost-at-icml-2017"},{"from":"\/news#version_0_2","to":"\/news\/version-0-2-released"},{"from":"\/news#version_0_3","to":"\/news\/extremely-fast-learning-on-gpu-has-arrived"},{"from":"\/news#version_0_6","to":"\/news\/best-in-class-inference-and-a-ton-of-speedups"},{"from":"\/news#gtc2018","to":"\/news\/catboost-on-gpu-talk-at-gtc-2018"},{"from":"\/news#version_0_8","to":"\/news\/new-ways-to-explore-your-data"}],"social":[{"socialNetworkName":"telegram","link":"https:\/\/t.me\/catboost_en","target":"_blank"},{"socialNetworkName":"twitter","link":"https:\/\/twitter.com\/CatBoostML","target":"_blank"}]},"nonce":"DqdhRCAqFwIwcRU8FqSbig==","staticHost":"\/\/yastatic.net\/s3\/distribution\/promo\/promo-catboost-2017\/v0.3.2","baseUrl":"","url":"https:\/\/catboost.ai","isMobile":false,"browser":"Chrome","bundle":"index","api":{"captcha":"\/captcha\/","subscribe":"\/subscribe\/","news":"\/api\/news\/"}});</script>
</body>
</html>
Sitemap
Кол-во: 0
XML-карта сайта для поисковиков
?
Sitemap.xml помогает поисковику быстрее находить и индексировать страницы. Особенно важен для крупных сайтов и новых страниц, на которые ещё нет входящих ссылок.
Robots.txt не содержит ссылку на карту сайта. Рекомендуется добавить карту сайта и указать ссылку на нее в robots.txt.
Внутренние ссылки
Кол-во: 38
Ссылки на другие страницы своего сайта
?
Внутренние ссылки распределяют ссылочный вес между страницами и помогают поисковику обходить сайт. Пустые анкоры и ссылки на запрещённые robots.txt страницы — типичные ошибки.
Внутренних ссылок на странице 38 оптимально.
Внутренние ссылки не запрещены к индексации в robots.txt.
На странице присутствуют изображения 1.
Показать внутренние ссылки
| Url | Анкор | Состояние |
|---|---|---|
| /en/concepts/installation |
<div class="dc-toc-item__text dc-toc-item__text_clicable dc-toc-item__text-block"><span>Overview<!-- --> </span></div>
|
|
| /en/concepts/parameter-tuning |
<div class="dc-toc-item__text dc-toc-item__text_clicable dc-toc-item__text-block"><span>Parameter tuning<!-- --> </span></div>
|
|
| /en/concepts/speed-up-training |
<div class="dc-toc-item__text dc-toc-item__text_clicable dc-toc-item__text-block"><span>Speeding up the training<!-- --> </span></div>
|
|
| /en/concepts/faq |
<div class="dc-toc-item__text dc-toc-item__text_clicable dc-toc-item__text-block"><span>FAQ<!-- --> </span></div>
|
|
| /en/concepts/development-and-contributions |
<div class="dc-toc-item__text dc-toc-item__text_clicable dc-toc-item__text-block"><span>Development and contributions<!-- --> </span></div>
|
|
| /en/concepts/contacts |
<div class="dc-toc-item__text dc-toc-item__text_clicable dc-toc-item__text-block"><span>Contacts<!-- --> </span></div>
|
|
| /en/features/training |
Training
|
|
| /en/features/training-on-gpu |
Training on GPU
|
|
| /en/concepts/python-reference_train |
Python train function
|
|
| /en/features/cross-validation |
Cross-validation
|
|
| /en/features/overfitting-detector-desc |
Overfitting detector
|
|
| /en/features/proceed-training |
Pre-trained data
|
|
| /en/features/categorical-features |
Categorical features
|
|
| /en/features/text-features |
Text features
|
|
| /en/features/embeddings-features |
Embeddings features
|
|
| /en/features/prediction |
Regular prediction
|
|
| /en/concepts/c-plus-plus-api |
С and C++
|
|
| /en/concepts/java-package |
Java
|
|
| /en/concepts/apply-node-js |
Node.js
|
|
| /en/concepts/apply-rust |
Rust
|
|
| /en/features/eval-metrics |
Calculate metrics
|
|
| /en/features/staged-prediction |
Staged prediction
|
|
| /en/features/catboost-with-clickhouse |
Applying the model in ClickHouse
|
|
| /en/features/feature-importances-calculation |
Feature importances
|
|
| /en/features/object-importances-calcution |
Object importances
|
|
| /en/features/loss-functions-desc |
Implemented metrics
|
|
| /en/features/custom-loss-functions |
User-defined metrics
|
|
| /en/features/snapshots |
Recovery
|
|
| /en/features/visualization_jupyter-notebook |
Jupyter Notebook
|
|
| /en/features/visualization_tensorboard |
TensorBoard
|
|
| /en/concepts/export-coreml |
CoreML
|
|
| /en/features/export-model-to-python |
Python or C++
|
|
| /en/features/export-model-to-json |
JSON
|
|
| /en/concepts/apply-onnx-ml |
ONNX
|
|
| /en/concepts/apply-pmml |
PMML
|
|
| /en/concepts/tutorials |
Tutorials
|
|
| /en/concepts/educational-materials-papers |
Reference papers
|
|
| /en/concepts/educational-materials-videos |
Videos
|
|
Внешние ссылки
Не найдено
Ссылки на сторонние сайты
?
Исходящие внешние ссылки передают часть ссылочного веса на чужие сайты. Ссылки на авторитетные ресурсы безопасны; ссылки на мусорные сайты могут навредить репутации страницы.
Внешних ссылок на странице не найдено.
Конкуренты Готовность: 0%
Конкуренты в Яндексе
Кол-во: 0
Топ сайтов-конкурентов в Яндексе
?
Сайты, чаще всего появляющиеся в ТОПе Яндекса по запросам из семантического ядра этой страницы.
Мы не нашли у вас конкурентов в Яндексе. Сайт или очень молодой или плохо продвигается.
Конкурентов в ТОП-10 Яндекса не нашлось.
Конкуренты в Google
Кол-во: 0
Топ сайтов-конкурентов в Google
?
Сайты, чаще всего появляющиеся в ТОПе Google по запросам из семантического ядра этой страницы.
Конкуренты в Google тоже не найдены. Займитесь продвижением сайта!
Конкурентов в ТОП-10 Google не нашлось.
ЗоЗПП: права потребителей Готовность: 100%
Нарушения
Не выявлены
Признаков дистанционной продажи товаров (интернет-магазина) не обнаружено — требования ЗоЗПП о раскрытии информации продавца к сайту не применяются. Нарушений нет.
ФЗ-149: рекомендательные технологии Готовность: 100%
Нарушения
Не выявлены
Рекомендательные блоки («с этим покупают», «похожие товары» и т.п.) на сайте не обнаружены — требования ст. 10.7 ФЗ-149 к сайту не применяются. Нарушений нет.
ФЗ-38: реклама Готовность: 100%
Нарушения
Не выявлены
Рекламных тематик с обязательными оговорками (медицина, БАД, кредиты и займы, новостройки) на сайте не обнаружено. Нарушений нет.
ФЗ-436: защита детей Готовность: 100%
Нарушения
Не выявлены
Признаков информационной продукции (новости, видео, книги, игры, курсы) не обнаружено — обязательная возрастная маркировка по ФЗ-436 сайту не требуется. Нарушений нет.
Вердикт
Страница https://catboost.ai/docs готова к продвижению на 74%. Чтобы еще улучшить страницу и попасть на первые места поисковой выдачи необходимо:
Исправьте ошибки в мета-тегах.
Исправьте ошибки оптимизации.
Поделитесь с друзьями: