pandas.testing.assert_index_equal doesn't produce results as expected #342
Description
Tests is not working on arm64 and aarch64:
lightweight_mmm/preprocessing_test.py::PreprocessingTest::test_check_data_quality_with_extra_features_national_data
lightweight_mmm/preprocessing_test.py::PreprocessingTest::test_compute_correlations_returns_expected_values_national_data
E DataFrame.iloc[:, 0] (column name="feature_0") values are different (20.0 %)
E [index]: [feature_0, feature_1, feature_2, feature_3, target]
E [left]: [1.0, 0.8338138461112976, 0.6024458408355713, 9.895054553510363e-09, 0.8184618949890137]
E [right]: [1.0, 0.83381, 0.60245, nan, 0.81847]
E At positional index 3, first diff: 9.895054553510363e-09 != nan
For some reason, float numbers have different precision.
There can be numpy and/or pandas incompatibilities or simply a bug in the source code.