Function InitializeSamplingStructures

catboost/libs/train_lib/train_model.cpp:312–342

}

static void InitializeSamplingStructures(
    const TTrainingDataProviders& data,
    TLearnContext* ctx) {

    const bool isPairwiseScoring = IsPairwiseScoring(ctx->Params.LossFunctionDescription->GetLossFunction());
    const int defaultCalcStatsObjBlockSize = static_cast<int>(ctx->Params.ObliviousTreeOptions->DevScoreCalcObjBlockSize);

    if (ctx->UseTreeLevelCaching()) {
        ctx->SmallestSplitSideDocs.Create(
            ctx->LearnProgress->Folds,
            isPairwiseScoring,
            data.EstimatedObjectsData.GetFeatureCount() != 0,
            defaultCalcStatsObjBlockSize
        );
        ctx->PrevTreeLevelStats.Create(
            ctx->LearnProgress->Folds,
            CountNonCtrBuckets(
                *data.Learn->ObjectsData->GetFeaturesLayout(),
                *data.Learn->ObjectsData->GetQuantizedFeaturesInfo(),
                ctx->Params.CatFeatureParams->OneHotMaxSize),
            static_cast<int>(ctx->Params.ObliviousTreeOptions->MaxDepth)
        );
    }
    ctx->SampledDocs.Create(
        ctx->LearnProgress->Folds,
        isPairwiseScoring,
        data.EstimatedObjectsData.GetFeatureCount() != 0,
        defaultCalcStatsObjBlockSize,
        GetBernoulliSampleRate(ctx->Params.ObliviousTreeOptions->BootstrapConfig)
    ); // TODO(espetrov): create only if sample rate < 1
}

static void LogThatStoppingOccured(const TErrorTracker& errorTracker) {
    CATBOOST_NOTICE_LOG << "Stopped by overfitting detector "