PyCon Belarus 2019, Minsk
Data Science Track
CatBoost: gradient boosting for classification, regression, and ranking – Anna Veronika Dorogush
CatBoost by Yandex is a library of gradient boosting. It is publicly available.
The main features of this library: it allows you to work effectively with categorical data, gives increased accuracy due to methods of dealing with retraining, realizes the ability to count model values for time-critical services quickly, and also gives the opportunity to train models on big data.
In the talk we will explain what gradient boasting is and why it is needed, will highlight the main features of the library and tell you what it is important to know and use while training gradient boosting models.
Video Partner – Wargaming (https://wargaming.com/ru/)