41 / 62 External Models CatBoost 43 / 62 Gradient Boosting General advantages › Best solution for heterogeneous data › Works well for small data › Easy to use CatBoost advantages › Good quality for Prediction time 52 / 62 Applying CatBoost models in ClickHouse CatBoost models in ClickHouse Steps to do: › Train model and save it as my_favorite_model.bin › Build CatBoost evaluation library. Follow the the instruction at https://catboost.ai/docs/concepts/c-plus-plus-api_dynamic-c-pluplus-wrapper.html You need to get libcatboostmodel.so › Update ClickHouse configuration file cat /etc/clickhouse-server/conf

41 / 62 External Models CatBoost 43 / 62 Gradient Boosting General advantages › Best solution for heterogeneous data › Works well for small data › Easy to use CatBoost advantages › Good quality for Prediction time 52 / 62 Applying CatBoost models in ClickHouse CatBoost models in ClickHouse Steps to do: › Train model and save it as my_favorite_model.bin › Build CatBoost evaluation library. Follow the the instruction at https://catboost.ai/docs/concepts/c-plus-plus-api_dynamic-c-pluplus-wrapper.html You need to get libcatboostmodel.so › Update ClickHouse configuration file cat /etc/clickhouse-server/conf