scikit-learn 数据处理之特征工程

1. 数据预处理

1. sklearn.feature_extraction
2. sklearn.impute
3. sklearn.preprocessing
4. pandas

import pandas as pd
# 1. load data
pdata_frame = pd.read_csv("file_path.csv", index_col=0)
print(pdata_frame.info)
# 2. fill na data
col_median = pdata_frame .loc[:, "col1"].median()
pdata_frame .loc[:, "col1"] = pdata_frame .loc[:, "col1"].fillna(col_median )
# 3. fill singular data
cols_median = pdata_frame .loc[:, "col2"].median()
pdata_frame .loc[pdata_frame ["col2"] < 0, "col2"] = cols_median
# 4. delete columns
data = pdata_frame .drop(["col3", "col5"], axis=1)
# 5. scaler
scaler = StandardScaler()
scaler.fit(data)
x_train = scaler.transform(data)

2. 特征降维

• sklearn.feature_selection
• sklearn.decomposition