Function CatBoostPoolSlice_R

catboost/R-package/src/catboostr.cpp:495–594

}

EXPORT_FUNCTION CatBoostPoolSlice_R(SEXP poolParam, SEXP sizeParam, SEXP offsetParam) {
    SEXP result = NULL;
    size_t size, offset;
    R_API_BEGIN();
    size = static_cast<size_t>(asInteger(sizeParam));
    offset = static_cast<size_t>(asInteger(offsetParam));
    TPoolHandle pool = static_cast<TPoolHandle>(R_ExternalPtrAddr(poolParam));
    const TRawObjectsDataProvider* rawObjectsData
        = dynamic_cast<const TRawObjectsDataProvider*>(pool->ObjectsData.Get());
    CB_ENSURE(rawObjectsData, "Cannot Slice quantized features data");

    const auto& featuresLayout = *(rawObjectsData->GetFeaturesLayout());

    CB_ENSURE(
        featuresLayout.GetExternalFeatureCount() == featuresLayout.GetFloatFeatureCount(),
        "Dataset slicing error: non-numeric features present, slicing datasets with categorical, text or embedding features is not supported"
    );

    result = PROTECT(allocVector(VECSXP, size));
    ui32 featureCount = pool->MetaInfo.GetFeatureCount();
    auto target = pool->RawTargetData.GetTarget();
    const auto& weights = pool->RawTargetData.GetWeights();


    const size_t sliceEnd = std::min((size_t)pool->GetObjectCount(), offset + size);

    TRangesSubset<ui32>::TBlocks subsetBlocks = { TSubsetBlock<ui32>(TIndexRange<ui32>(offset, sliceEnd), 0) };

    TObjectsGroupingSubset objectsGroupingSubset = GetGroupingSubsetFromObjectsSubset(
        rawObjectsData->GetObjectsGrouping(),
        TArraySubsetIndexing<ui32>(TRangesSubset<ui32>(subsetBlocks[0].GetSize(), std::move(subsetBlocks))),
        EObjectsOrder::Ordered
    );

    TObjectsDataProviderPtr sliceObjectsData = rawObjectsData->GetSubset(
        objectsGroupingSubset,
        GetMonopolisticFreeCpuRam(),
        &NPar::LocalExecutor()
    );

    const TRawObjectsDataProvider& sliceRawObjectsData
        = dynamic_cast<const TRawObjectsDataProvider&>(*sliceObjectsData);

    TVector<double*> rows;
    const auto targetCount = pool->MetaInfo.TargetCount;

    for (size_t i = offset; i < sliceEnd; ++i) {
        ui32 featureCount = pool->MetaInfo.GetFeatureCount();
        SEXP row = PROTECT(allocVector(REALSXP, featureCount + targetCount + 1));
        REAL(row)[targetCount] = weights[i];
        rows.push_back(REAL(row));
        SET_VECTOR_ELT(result, i - offset, row);
    }

    for (auto targetIdx : xrange(targetCount)) {
        if (const ITypedSequencePtr<float>* typedSequence
                = std::get_if<ITypedSequencePtr<float>>(&((*target)[targetIdx])))
        {