| 76 | } |
| 77 | |
| 78 | NCB::TDataMetaInfo GetDataMetaInfo( |
| 79 | const NCB::TQuantizedPool& pool, |
| 80 | bool hasAdditionalGroupWeight, |
| 81 | bool hasTimestamps, |
| 82 | bool hasPairs, |
| 83 | bool forceUnitAutoPairWeights, |
| 84 | TMaybe<ui32> baselineCount, |
| 85 | const NCB::TPathWithScheme& featureNamesPath, |
| 86 | const NCB::TPathWithScheme& poolMetaInfoPath |
| 87 | ) { |
| 88 | const size_t columnsCount = pool.ColumnIndexToLocalIndex.size(); |
| 89 | NCB::TDataColumnsMetaInfo dataColumnsMetaInfo; |
| 90 | dataColumnsMetaInfo.Columns.resize(columnsCount); |
| 91 | |
| 92 | bool hasTargets = false; |
| 93 | |
| 94 | for (const auto [columnIndex, localIndex] : pool.ColumnIndexToLocalIndex) { |
| 95 | const auto columnType = pool.ColumnTypes[localIndex]; |
| 96 | if (columnType == EColumn::Label) { |
| 97 | hasTargets = true; |
| 98 | } |
| 99 | dataColumnsMetaInfo.Columns[columnIndex].Type = pool.ColumnTypes[localIndex]; |
| 100 | dataColumnsMetaInfo.Columns[columnIndex].Id = pool.ColumnNames[localIndex]; |
| 101 | } |
| 102 | |
| 103 | NCB::ERawTargetType targetType; |
| 104 | if (hasTargets) { |
| 105 | if (pool.QuantizationSchema.IntegerClassLabelsSize()) { |
| 106 | targetType = NCB::ERawTargetType::Integer; |
| 107 | } else if (pool.QuantizationSchema.FloatClassLabelsSize()) { |
| 108 | targetType = NCB::ERawTargetType::Float; |
| 109 | } else if (pool.QuantizationSchema.ClassNamesSize()) { |
| 110 | targetType = NCB::ERawTargetType::String; |
| 111 | } else { |
| 112 | targetType = NCB::ERawTargetType::Float; |
| 113 | } |
| 114 | } else { |
| 115 | targetType = NCB::ERawTargetType::None; |
| 116 | } |
| 117 | |
| 118 | const TVector<TString> featureNames = NCB::GetFeatureNames( |
| 119 | dataColumnsMetaInfo, |
| 120 | /*headerColumns*/ Nothing(), |
| 121 | featureNamesPath |
| 122 | ); |
| 123 | |
| 124 | const auto poolMetaInfoOptions = NCatboostOptions::LoadPoolMetaInfoOptions(poolMetaInfoPath); |
| 125 | |
| 126 | NCB::TDataMetaInfo metaInfo( |
| 127 | std::move(dataColumnsMetaInfo), |
| 128 | targetType, |
| 129 | hasAdditionalGroupWeight, |
| 130 | hasTimestamps, |
| 131 | hasPairs, |
| 132 | /*hasGraph*/ false, |
| 133 | /*loadSampleIds*/ false, |
| 134 | forceUnitAutoPairWeights, |
| 135 | baselineCount, |
no test coverage detected