CatBoost — gradient boosting training on the large data volumes

Day 1 /  / Track 3  /  For practicing engineers

CatBoost is a gradient boosting library; Yandex made it open-source.

The main features of the library are as follows: it allows effective work with the categorical data, improves accuracy by using overfitting control methods, allows quick calculating of the values of the model for the time-critical services, and gives an opportunity to train models on the large data volumes. In this session we are going to talk briefly about the meaning and functions of the gradient boosting, cover the library's main features, and dwell on boosting training on the large data volumes.