-
Notifications
You must be signed in to change notification settings - Fork 44
/
correlation_similarity.py
143 lines (114 loc) · 5.17 KB
/
correlation_similarity.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
"""Correlation Similarity Metric."""
import warnings
import numpy as np
import pandas as pd
from scipy.stats import pearsonr, spearmanr
from sdmetrics.column_pairs.base import ColumnPairsMetric
from sdmetrics.goal import Goal
from sdmetrics.utils import is_datetime
from sdmetrics.warnings import ConstantInputWarning
class CorrelationSimilarity(ColumnPairsMetric):
"""Correlation similarity metric.
Attributes:
name (str):
Name to use when reports about this metric are printed.
goal (sdmetrics.goal.Goal):
The goal of this metric.
min_value (Union[float, tuple[float]]):
Minimum value or values that this metric can take.
max_value (Union[float, tuple[float]]):
Maximum value or values that this metric can take.
"""
name = 'CorrelationSimilarity'
goal = Goal.MAXIMIZE
min_value = 0.0
max_value = 1.0
@staticmethod
def _generate_warning_msg(columns, prefix, warning_messages):
if len(columns) > 1:
cols = ', '.join(columns)
warning_messages.append(
f"The {prefix} in columns '{cols}' contain a constant value. "
'Correlation is undefined for constant data.'
)
elif len(columns):
warning_messages.append(
f"The {prefix} in column '{columns[0]}' contains a constant value. "
'Correlation is undefined for constant data.'
)
@classmethod
def compute_breakdown(cls, real_data, synthetic_data, coefficient='Pearson'):
"""Compare the breakdown of correlation similarity of two continuous columns.
Args:
real_data (Union[numpy.ndarray, pandas.Series]):
The values from the real dataset.
synthetic_data (Union[numpy.ndarray, pandas.Series]):
The values from the synthetic dataset.
Returns:
dict:
A dict containing the score, and the real and synthetic metric values.
"""
real_data = real_data.copy()
synthetic_data = synthetic_data.copy()
if not isinstance(real_data, pd.DataFrame):
real_data = pd.DataFrame(real_data)
synthetic_data = pd.DataFrame(synthetic_data)
if (real_data.nunique() == 1).any() or (synthetic_data.nunique() == 1).any():
warning_messages = []
real_columns = list(real_data.loc[:, real_data.nunique() == 1].columns)
synthetic_columns = list(synthetic_data.loc[:, synthetic_data.nunique() == 1].columns)
cls._generate_warning_msg(real_columns, 'real data', warning_messages)
cls._generate_warning_msg(synthetic_columns, 'synthetic data', warning_messages)
for msg in warning_messages:
warnings.warn(ConstantInputWarning(msg))
return {'score': np.nan}
column1, column2 = real_data.columns[:2]
real_data = real_data[[column1, column2]].dropna()
synthetic_data = synthetic_data[[column1, column2]].dropna()
if is_datetime(real_data[column1]):
real_data[column1] = pd.to_numeric(real_data[column1])
synthetic_data[column1] = pd.to_numeric(synthetic_data[column1])
if is_datetime(real_data[column2]):
real_data[column2] = pd.to_numeric(real_data[column2])
synthetic_data[column2] = pd.to_numeric(synthetic_data[column2])
correlation_fn = None
if coefficient == 'Pearson':
correlation_fn = pearsonr
elif coefficient == 'Spearman':
correlation_fn = spearmanr
else:
raise ValueError(f'requested coefficient {coefficient} is not valid. '
'Please choose either Pearson or Spearman.')
correlation_real, _ = correlation_fn(real_data[column1], real_data[column2])
correlation_synthetic, _ = correlation_fn(synthetic_data[column1], synthetic_data[column2])
if np.isnan(correlation_real) or np.isnan(correlation_synthetic):
return {'score': np.nan}
return {
'score': 1 - abs(correlation_real - correlation_synthetic) / 2,
'real': correlation_real,
'synthetic': correlation_synthetic,
}
@classmethod
def compute(cls, real_data, synthetic_data, coefficient='Pearson'):
"""Compare the correlation similarity of two continuous columns.
Args:
real_data (Union[numpy.ndarray, pandas.Series]):
The values from the real dataset.
synthetic_data (Union[numpy.ndarray, pandas.Series]):
The values from the synthetic dataset.
Returns:
float:
The correlation similarity of the two columns.
"""
return cls.compute_breakdown(real_data, synthetic_data, coefficient)['score']
@classmethod
def normalize(cls, raw_score):
"""Return the `raw_score` as is, since it is already normalized.
Args:
raw_score (float):
The value of the metric from `compute`.
Returns:
float:
The normalized value of the metric
"""
return super().normalize(raw_score)