def statistic_derivation(table, group_by=None, **params):
check_required_parameters(_statistic_derivation, params, ['table'])
params = get_default_from_parameters_if_required(params,_statistic_derivation)
param_validation_check = [all_elements_from_to(params, 0, 100, 'percentile_amounts'),
all_elements_from_under(params, 0, 0.5, 'trimmed_mean_amounts')]
validate(*param_validation_check)
if group_by is not None:
return _function_by_group(_statistic_derivation, table, group_by=group_by, **params)
else:
return _statistic_derivation(table, **params)
def statistic_summary(table, group_by=None, **params):
check_required_parameters(_statistic_summary, params, ['table'])
params = get_default_from_parameters_if_required(params,
_statistic_summary)
param_validation_check = [
all_elements_from_to(params, 0, 100, 'percentile_amounts'),
all_elements_from_under(params, 0, 0.5, 'trimmed_mean_amounts')
]
validate(*param_validation_check)
column_indices = []
for i in params['input_cols']:
column_indices.append(table.columns.get_loc(i))
params['column_indices'] = column_indices
columns = ['column_name'] + params['statistics'].copy()
columns = ['num_of_row' if x == 'nrow' else x for x in columns]
if 'percentile' in columns:
columns.remove('percentile')
if params['percentile_amounts'] is not None:
for pa in _unique_list(params['percentile_amounts']):
columns.append('percentile_{}'.format(_amounts_colname(pa)))
if 'trimmed_mean' in columns:
columns.remove('trimmed_mean')
if params['trimmed_mean_amounts'] is not None:
for ta in _unique_list(params['trimmed_mean_amounts']):
columns.append('trimmed_mean_{}'.format(_amounts_colname(ta)))
if 'mode' in columns:
columns.remove('mode')
columns.append('mode')
if group_by is not None:
return _function_by_group2(_statistic_summary,
table,
columns=columns,
group_by=group_by,
**params)
else:
tmp_table = table.values
if 'workers' in params:
del params['workers']
result = _statistic_summary(tmp_table, **params)
result['out_table'] = pd.DataFrame(result['out_table'],
columns=columns)
return result
def statistic_summary(table, group_by=None, **params):
check_required_parameters(_statistic_summary_list, params, ['table'])
params = get_default_from_parameters_if_required(
params, _statistic_summary_original)
param_validation_check = [
all_elements_from_to(params, 0, 100, 'percentile_amounts'),
all_elements_from_under(params, 0, 0.5, 'trimmed_mean_amounts')
]
validate(*param_validation_check)
if group_by is None:
return _statistic_summary_original(table, **params)
if True in pd.isnull(table[group_by]).values:
group_by_unicode = [str(i) + '\u0003' for i in group_by]
table[group_by_unicode] = table[group_by].fillna('\u0003')
group_by = group_by_unicode
params1 = dict()
params1['input_cols'] = params['input_cols']
params2 = dict()
params1['statistics'] = []
params2['statistics'] = []
for st in params['statistics']:
if st in [
'max', 'min', 'range', 'sum', 'avg', 'variance', 'stddev',
'nrow', 'num_of_value', 'null_count', 'median'
]:
params1['statistics'].append(st)
else:
params2['statistics'].append(st)
for st in params.keys():
if 'percentile' == st:
params2['percentile'] = params['percentile']
if 'trimmed_mean' == st:
params2['trimmed_mean'] = params['trimmed_mean']
if 'percentile_amounts' == st:
params2['percentile_amounts'] = params['percentile_amounts']
if 'trimmed_mean_amounts' == st:
params2['trimmed_mean_amounts'] = params['trimmed_mean_amounts']
if params1['statistics']:
result1 = _statistic_summary_groupby(table,
group_by=group_by,
**params1)
result1.index = result1[group_by + ['column_name']]
else:
result1 = None
if params2['statistics']:
params2['input_cols'] = params['input_cols']
column_indices = []
for i in params2['input_cols']:
column_indices.append(table.columns.get_loc(i))
params2['column_indices'] = column_indices
columns = ['column_name'] + params2['statistics'].copy()
columns = ['num_of_row' if x == 'nrow' else x for x in columns]
if 'percentile' in columns:
columns.remove('percentile')
if params2['percentile_amounts'] is not None:
for pa in _unique_list(params2['percentile_amounts']):
columns.append('percentile_{}'.format(
_amounts_colname(pa)))
if 'trimmed_mean' in columns:
columns.remove('trimmed_mean')
if params2['trimmed_mean_amounts'] is not None:
for ta in _unique_list(params2['trimmed_mean_amounts']):
columns.append('trimmed_mean_{}'.format(
_amounts_colname(ta)))
if 'mode' in columns:
columns.remove('mode')
columns.append('mode')
result2 = _function_by_group2(_statistic_summary_list,
table,
columns=columns,
group_by=group_by,
**params2)['out_table']
result2.index = result2[group_by + ['column_name']]
if result1 is not None:
result2 = result2[[
i for i in result2.columns
if i not in group_by + ['column_name']
]]
# Update sort parameter after upgrading pandas.
# result = pd.concat([result2,result1], axis=1, sort = False).reset_index(drop = True)
result = pd.concat([result2, result1],
axis=1).reset_index(drop=True)
else:
result = result2
else:
groups = table[group_by].drop_duplicates().values
result2 = []
for i in groups:
for j in params['input_cols']:
result2.append(list(i) + [j])
result2 = pd.DataFrame(result2, columns=group_by + ['column_name'])
result2.index = result2[group_by + ['column_name']]
result1 = result1[[
i for i in result1.columns if i not in group_by + ['column_name']
]]
# Update sort parameter after upgrading pandas.
# result = pd.concat([result2,result1], axis=1, sort=False).reset_index(drop = True)
result = pd.concat([result2, result1], axis=1).reset_index(drop=True)
columns = ['column_name'] + params['statistics'].copy()
if 'percentile' in columns:
columns.remove('percentile')
if params['percentile_amounts'] is not None:
for pa in _unique_list(params['percentile_amounts']):
columns.append('percentile_{}'.format(_amounts_colname(pa)))
if 'trimmed_mean' in columns:
columns.remove('trimmed_mean')
if params['trimmed_mean_amounts'] is not None:
for ta in _unique_list(params['trimmed_mean_amounts']):
columns.append('trimmed_mean_{}'.format(_amounts_colname(ta)))
if 'mode' in columns:
columns.remove('mode')
columns.append('mode')
result = result[group_by + columns]
if '\u0003' in result.values:
result = result.replace('\u0003', np.nan)
tmp_col = result.columns
result.columns = [_remove_unicode(i) for i in tmp_col]
return {'out_table': result}