/
feature_engineering.py
45 lines (29 loc) · 1.33 KB
/
feature_engineering.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
from dataclasses import dataclass
from typing import List
import pandas as pd
from dataclass import DataInfo, TrainDataInfo
@dataclass
class TrainFeatureEngineeringResult:
X: pd.DataFrame
Y: pd.DataFrame
feature_columns: List[str]
def get_numerical_cols(data: pd.DataFrame) -> pd.Index:
numerical_columns: pd.Index = data.select_dtypes(exclude='object').columns
return numerical_columns
def verify(data: pd.DataFrame, expected_info: DataInfo):
assert data.columns.size == len(expected_info.column_dict)
assert len(data) == expected_info.size
def train_feature_engineering(data: pd.DataFrame, expected_info: TrainDataInfo) -> TrainFeatureEngineeringResult:
feature_cols = [col for col in get_numerical_cols(data) if
col != expected_info.label_column.name and not expected_info.column_dict[col].exclude]
print(feature_cols)
X = data[feature_cols]
X = X.fillna(-1)
Y = data[expected_info.label_column.name]
return TrainFeatureEngineeringResult(X, Y, feature_cols)
def predict_feature_engineering(data: pd.DataFrame,
train_feature_engineering_result: TrainFeatureEngineeringResult) -> pd.DataFrame:
X = data[train_feature_engineering_result.feature_columns]
X = X.fillna(-1)
# todo: sync with train feature engineering
return X