def test_dataset_analysis_report(self):
da = DatasetAnalysis(REGRESSION,
class_count_threshold=10,
outliers_n_sigma=(3, 5),
nan_threshold_per_col=(0.1, 0.5, 0.9),
nan_threshold_per_sample=(0.1, 0.9))
report = da.dataset_analysis_report(dataset=self.dataset)
pprint(report.to_dict(printable=True))
pprint(report.to_dict_meaning())
pprint(report.plot())
def create_dataset_analysis_report(
cls,
train: DMD,
is_classification,
test: DMD = None,
dataset_analysis: DatasetAnalysis = None,
**kwargs) -> DatasetAnalysisReport:
"""
Create dataset analysis report by analyzing train data
Args:
train - train data
test - test data, if available
is_classification - whether the target is categorical.
Returns:
dataset analysis report
"""
dataset_analysis = dataset_analysis or DatasetAnalysis(
problem_Type=CLASSIFICATION if is_classification else REGRESSION)
report = dataset_analysis.dataset_analysis_report(train=train,
test=test)
return report
def create_dataset_analysis_report(cls, train: DMD, is_classification,
**kwargs) -> DatasetAnalysisReport:
da = DatasetAnalysis(
problem_Type=CLASSIFICATION if is_classification else REGRESSION)
report = da.dataset_analysis_report(dataset=train)
return report