scikit-learn-intelex
scikit-learn-intelex copied to clipboard
intel
IncrementalLinearRegression interface
Changes proposed in this pull request:
- IncrementalLinearRegression added to onedal/sklearnex
- Contains changes from https://github.com/intel/scikit-learn-intelex/pull/1713
@olegkkruglov please rebase your branch.
It make sense think first about general online algos test coverage. Especially that estimators that we are not patching. I expect, that we will probably add some sklearn like tests. Some basic pikling, n_jobs, testing via different dataframes and device contexts.
Generally it is required first update primitives in case of new online algos. For example, this IncrementalLinReg will not work with config_context for device offloading. etc.
Ideally this should inherit sklearns' RegressorMixin too, just like BaseEstimator.
New PR introduced to manage the issues with test_patching and _utils.py. We need to talk to @Alexsandruss about n_jobs testing and how that interacts with partial_fit
conformance internal CI fails: FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_estimators_pickle] FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_fit2d_1sample] FAILED tests/test_common.py::test_estimators[LinearRegression()-check_estimators_unfitted]
conformance internal CI fails: FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_estimators_pickle] FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_fit2d_1sample] FAILED tests/test_common.py::test_estimators[LinearRegression()-check_estimators_unfitted]
These still need to be resolved
With the fix in test_patching, it looks like there is some sort of data conversion issue with fit (not in float types).
![azure-pipelines[bot] avatar](https://avatars.githubusercontent.com/in/9426?v=4 "Published by a pub.dev verified publisher"){kind=small}
Internal CI conformance fails including but not limited to: FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_estimators_nan_inf] FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_regressors_train] FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_regressors_train(readonly_memmap=True)] FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_regressors_train(readonly_memmap=True,X_dtype=float32)] FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_regressor_data_not_an_array] FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_supervised_y_2d] FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_estimators_unfitted] FAILED tests/test_common.py::test_estimators[IncrementalLinearRegression()-check_fit1d] FAILED tests/test_common.py::test_check_param_validation[IncrementalLinearRegression()]
https://ecmd.jf.intel.com/commander/link/jobDetails/jobs/eedd31b2-8f0f-f1ce-bd49-a4bf010d0e2e