diff options
-rw-r--r-- | lib/parameters.py | 10 |
1 files changed, 6 insertions, 4 deletions
diff --git a/lib/parameters.py b/lib/parameters.py index 45a577b..d0894e4 100644 --- a/lib/parameters.py +++ b/lib/parameters.py @@ -43,12 +43,10 @@ def distinct_param_values(param_tuples): def _depends_on_param(corr_param, std_param, std_lut, threshold=0.5): - # if self.use_corrcoef: - if False: - return corr_param > 0.1 - elif std_param == 0: + if std_param == 0: # In general, std_param_lut < std_by_param. So, if std_by_param == 0, std_param_lut == 0 follows. # This means that the variation of param does not affect the model quality -> no influence + # assert std_lut == 0 return False return std_lut / std_param < threshold @@ -1160,6 +1158,10 @@ class ModelAttribute: relevance_threshold = float(os.getenv("DFATOOL_PARAM_RELEVANCE_THRESHOLD", 0.5)) + logger.debug( + f"build_dtree(threshold={threshold}, relevance_threshold={relevance_threshold})" + ) + self.model_function = self._build_dtree( parameters, data, |