def _pandas_value_ratio(
self,
batches: Dict[str, Batch],
execution_engine: PandasExecutionEngine,
metric_domain_kwargs: Dict,
metric_value_kwargs: Dict,
metrics: Dict,
runtime_configuration: dict = None,
):
"""Value Ratio MetricProvider Function, extracts nonnull count to use for obtaining the value ratio"""
# Column Extraction
series = execution_engine.get_domain_dataframe(
domain_kwargs=metric_domain_kwargs, batches=batches)
domain_metrics_lookup = get_domain_metrics_dict_by_name(
metrics=metrics, metric_domain_kwargs=metric_domain_kwargs)
nonnull_count = domain_metrics_lookup[
"column_values.nonnull.unexpected_count"]
wanted_value = metric_value_kwargs["value"]
# Checking that the wanted value is indeed in the value set itself
if wanted_value in series.value_counts():
value_count = series.value_counts()[wanted_value]
else:
value_count = 0
return value_count / nonnull_count
def _pandas_columns(
self,
batches: Dict[str, Batch],
execution_engine: PandasExecutionEngine,
metric_domain_kwargs: Dict,
metric_value_kwargs: Dict,
metrics: Dict,
runtime_configuration: dict = None,
):
"""Metric which returns all columns in a dataframe"""
df = execution_engine.get_domain_dataframe(
domain_kwargs=metric_domain_kwargs, batches=batches)
cols = df.columns
return cols.tolist()
def _pandas_column_a_greater_than_b(
self,
batches: Dict[str, Batch],
execution_engine: PandasExecutionEngine,
metric_domain_kwargs: Dict,
metric_value_kwargs: Dict,
metrics: Dict,
runtime_configuration: dict = None,
):
"""Metric which returns all columns in a dataframe"""
df = execution_engine.get_domain_dataframe(
domain_kwargs=metric_domain_kwargs, batches=batches)
# Initialization of necessary value kwargs
allow_cross_type_comparisons = None
parse_strings_as_datetimes = None
or_equal = None
column_A = df[metric_value_kwargs["column_A"]]
column_B = df[metric_value_kwargs["column_B"]]
# If value kwargs are given that could impact outcome, initializing them
if allow_cross_type_comparisons in metric_value_kwargs:
allow_cross_type_comparisons = metric_value_kwargs[
"allow_cross_type_comparisons"]
if parse_strings_as_datetimes in metric_value_kwargs:
parse_strings_as_datetimes = metric_value_kwargs[
"parse_strings_as_datetimes"]
if or_equal in metric_value_kwargs:
or_equal = metric_value_kwargs["or_equal"]
if allow_cross_type_comparisons:
column_A = column_A.apply(str)
column_B = column_B.apply(str)
if parse_strings_as_datetimes:
temp_column_A = column_A.map(parse)
temp_column_B = column_B.map(parse)
else:
temp_column_A = column_A
temp_column_B = column_B
if or_equal:
return temp_column_A >= temp_column_B
else:
return temp_column_A > temp_column_B
def _pandas_equal_columns(
self,
batches: Dict[str, Batch],
execution_engine: PandasExecutionEngine,
metric_domain_kwargs: Dict,
metric_value_kwargs: Dict,
metrics: Dict,
runtime_configuration: dict = None,
):
"""Metric which returns all columns in a dataframe"""
df = execution_engine.get_domain_dataframe(
domain_kwargs=metric_domain_kwargs, batches=batches)
column_A = df[metric_value_kwargs["column_A"]]
column_B = df[metric_value_kwargs["column_B"]]
return (column_A == column_B).any()
