| 255 | |
| 256 | template <class TClassId> |
| 257 | TVector<TVector<ui32>> SplitByClass( |
| 258 | const TObjectsGrouping& objectsGrouping, |
| 259 | TConstArrayRef<TClassId> objectClasses |
| 260 | ) { |
| 261 | CB_ENSURE(objectsGrouping.IsTrivial(), "Stratified split is not supported for data with groups"); |
| 262 | |
| 263 | const ui32 objectCount = objectsGrouping.GetObjectCount(); |
| 264 | |
| 265 | CheckDataSize( |
| 266 | objectClasses.size(), |
| 267 | (size_t)objectCount, |
| 268 | "objectClasses", |
| 269 | false, |
| 270 | "objects size", |
| 271 | true |
| 272 | ); |
| 273 | |
| 274 | TVector<std::pair<TClassId, ui32>> classWithObject; |
| 275 | classWithObject.reserve(objectCount); |
| 276 | for (ui32 i = 0; i < objectCount; ++i) { |
| 277 | classWithObject.emplace_back(objectClasses[i], i); |
| 278 | } |
| 279 | Sort(classWithObject.begin(), classWithObject.end()); |
| 280 | |
| 281 | TVector<TVector<ui32>> splittedByClass; |
| 282 | for (ui32 i = 0; i < classWithObject.size(); ++i) { |
| 283 | if (i == 0 || classWithObject[i].first != classWithObject[i - 1].first) { |
| 284 | splittedByClass.emplace_back(); |
| 285 | } |
| 286 | splittedByClass.back().push_back(classWithObject[i].second); |
| 287 | } |
| 288 | |
| 289 | return splittedByClass; |
| 290 | } |
| 291 | |
| 292 | template <class TClassId> |
| 293 | void StratifiedTrainTestSplit( |
no test coverage detected