Function GetPredictionDiff

catboost/libs/fstr/compare_documents.cpp:165–249

}

TVector<double> GetPredictionDiff(
    const TFullModel& model,
    const TObjectsDataProviderPtr objectsDataProvider,
    NPar::ILocalExecutor* localExecutor
) {
    CB_ENSURE(model.ModelTrees->GetDimensionsCount() == 1,  "Is not supported for multiclass");
    CB_ENSURE(objectsDataProvider->GetObjectCount() == 2, "PredictionDiff requires 2 documents for compare");
    CB_ENSURE(model.GetNumCatFeatures() == 0, "Models with categorical features are not supported");

    TVector<ui32> leafIdxes = CalcLeafIndexesMulti(model, objectsDataProvider, 0, 0);

    const auto* const rawObjectsData = dynamic_cast<const TRawObjectsDataProvider*>(objectsDataProvider.Get());

    const auto& binSplits = model.ModelTrees->GetBinFeatures();

    TVector<TVector<double>> floatFeatureValues(objectsDataProvider->GetObjectCount());
    for (size_t idx = 0; idx < objectsDataProvider->GetObjectCount(); ++idx) {
        floatFeatureValues[idx].resize(model.GetNumFloatFeatures());
    }

    TVector<TVector<ui32>> docBorders(objectsDataProvider->GetObjectCount());
    TVector<ui32> borderIdxForSplit(binSplits.size(), std::numeric_limits<ui32>::infinity());
    ui32 splitIdx = 0;
    for (const auto& feature : model.ModelTrees->GetFloatFeatures()) {
        TMaybeData<const TFloatValuesHolder*> featureData
            = rawObjectsData->GetFloatFeature(feature.Position.FlatIndex);

        if (const auto* arrayColumn = dynamic_cast<const TFloatArrayValuesHolder*>(*featureData)) {
            arrayColumn->GetData()->ParallelForEach(
                [&] (ui32 docId, float value) {
                    docBorders[docId].push_back(0);
                    for (const auto& border: feature.Borders) {
                        if (value > border) {
                            docBorders[docId].back()++;
                            floatFeatureValues[docId][feature.Position.FlatIndex] = value;
                        }
                    }
                },
                localExecutor
            );
        } else {
            CB_ENSURE_INTERNAL(false, "GetPredictionDiff: Unsupported column type");
        }
        if (splitIdx == binSplits.size() ||
            binSplits[splitIdx].Type != ESplitType::FloatFeature ||
            binSplits[splitIdx].FloatFeature.FloatFeature > feature.Position.Index
        ) {
            continue;
        }
        Y_ASSERT(binSplits[splitIdx].FloatFeature.FloatFeature >= feature.Position.Index);
        for (ui32 idx = 0; idx < feature.Borders.size() && binSplits[splitIdx].FloatFeature.FloatFeature == feature.Position.Index; ++idx) {
            if (abs(binSplits[splitIdx].FloatFeature.Split - feature.Borders[idx]) < 1e-15) {
                borderIdxForSplit[splitIdx] = idx;
                ++splitIdx;
            }
        }
    }