Python数据标准化归一化方法教程

用pandas和sklearn做Min-Max归一化、Z-score标准化，解决不同量纲字段（金额与数量）在模型、相似度计算中权重失衡的问题。

场景痛点

做客户分群、相似推荐、回归预测时，金额列动辄几万、数量列只有个位数，不标准化的话金额会主导计算结果。Excel里手动算 (x-min)/(max-min) 每列拖一遍，几十列就疯了。用pandas一行公式或sklearn的Scaler，几行搞定。

用到的库

pip install pandas scikit-learn

完整代码

import pandas as pd
from sklearn.preprocessing import MinMaxScaler, StandardScaler

def sample_df():
return pd.DataFrame({
"amount":  [1000, 5000, 10000, 20000, 3000],
"qty":     [1, 3, 5, 10, 2],
"credit":  [600, 650, 700, 800, 620],
})

def norm_demo():
df = sample_df()
print("原始：")
print(df)

# 1. Min-Max归一化：缩放到 [0,1]，用pandas手写
df["amount_minmax"] = (df["amount"] - df["amount"].min()) / (df["amount"].max() - df["amount"].min())

# 2. 用sklearn批量归一化多列
scaler = MinMaxScaler()
scaled = scaler.fit_transform(df[["amount", "qty", "credit"]])
df_scaled = pd.DataFrame(scaled, columns=["amount_m", "qty_m", "credit_m"])
print("\nMin-Max归一化后：")
print(df_scaled.round(3))

# 3. Z-score标准化：均值0，标准差1
std_scaler = StandardScaler()
std_arr = std_scaler.fit_transform(df[["amount", "qty", "credit"]])
df_std = pd.DataFrame(std_arr, columns=["amount_z", "qty_z", "credit_z"])
print("\nZ-score标准化后：")
print(df_std.round(3))

# 4. 反归一化：把模型结果还原回原始量纲
back = scaler.inverse_transform(df_scaled)
print("\n反归一化还原：")
print(pd.DataFrame(back, columns=["amount", "qty", "credit"]).round(2))

if __name__ == "__main__":
norm_demo()