XGboostRegressorBostonHousingNumericTest is failing

[zaccharieramzi]

The numeric test for xgboost regressor converter is failing with the following:

FAIL: test_boston_housing_simple_random_forest_regression (coremltools.test.test_boosted_trees_regression_numeric.XGboostRegressorBostonHousingNumericTest)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/zaccharieramzi/workspace/coremltools/coremltools/test/test_boosted_trees_regression_numeric.py", line 223, in test_boston_housing_simple_random_forest_regression
    self._train_convert_evaluate_assert(subsample = 0.5)
  File "/Users/zaccharieramzi/workspace/coremltools/coremltools/test/test_boosted_trees_regression_numeric.py", line 217, in _train_convert_evaluate_assert
    self._check_metrics(metrics, bt_params)
  File "/Users/zaccharieramzi/workspace/coremltools/coremltools/test/test_boosted_trees_regression_numeric.py", line 195, in _check_metrics
    msg = 'Failed case %s. Results %s' % (params, metrics))
AssertionError: 0.016184903142875712 != 0 within 0.0001 delta (0.016184903142875712 difference) : Failed case {}. Results {'samples': 506, 'rmse': 0.016184903142875712, 'max_error': 0.2887819620530898}

I am on mac OS 10.14.

[1duo]

I believe this issue is resolved with your PR https://github.com/apple/coremltools/pull/293, right? Feel free to re-open if it's not the case. Thanks for the fix BTW!

[zaccharieramzi]

I don't think so. I can't re-open so please do

[TobyRoseman]

The relevant file has moved. However we're still skipping that test. I get an Abort trap: 6 error when I try to run it.