| 42 | |
| 43 | |
| 44 | static TDataProviderPtr ReadDatasetForSampler( |
| 45 | const TDataProviderSampleParams& params, |
| 46 | bool loadSampleIds, |
| 47 | THolder<ILineDataReader>&& lineDataReader |
| 48 | ) { |
| 49 | const auto& datasetReadingParams = params.DatasetReadingParams; |
| 50 | |
| 51 | TVector<TColumn> columnsDescription; |
| 52 | TMaybe<size_t> sampleIdColumnIdx; |
| 53 | |
| 54 | ReadCDForSampler( |
| 55 | datasetReadingParams.ColumnarPoolFormatParams.CdFilePath, |
| 56 | params.OnlyFeaturesData, |
| 57 | loadSampleIds, |
| 58 | &columnsDescription, |
| 59 | &sampleIdColumnIdx |
| 60 | ); |
| 61 | |
| 62 | TVector<NJson::TJsonValue> classLabels = datasetReadingParams.ClassLabels; |
| 63 | |
| 64 | return ReadDataset( |
| 65 | std::move(lineDataReader), |
| 66 | /*pairsFilePath*/ TPathWithScheme(), |
| 67 | /*graphFilePath*/ TPathWithScheme(), |
| 68 | /*groupWeightsFilePath*/ TPathWithScheme(), |
| 69 | /*timestampsFilePath*/ TPathWithScheme(), |
| 70 | /*baselineFilePath*/ TPathWithScheme(), |
| 71 | datasetReadingParams.FeatureNamesPath, |
| 72 | datasetReadingParams.PoolMetaInfoPath, |
| 73 | datasetReadingParams.ColumnarPoolFormatParams.DsvFormat, |
| 74 | columnsDescription, |
| 75 | /*ignoredFeatures*/ {}, //TODO(akhropov): get unused features from the model |
| 76 | EObjectsOrder::Ordered, |
| 77 | loadSampleIds, |
| 78 | /*forceUnitAutoPairWeights*/ false, |
| 79 | &classLabels, |
| 80 | params.LocalExecutor |
| 81 | ); |
| 82 | } |
| 83 | |
| 84 | void DegroupDataset(TDataProvider* dataset) { |
| 85 | auto& metaInfo = dataset->MetaInfo; |
no test coverage detected