异常值处理、3sigma准则

import  pandas as pd


# 正态分布
# 3sigma准则 --->  
#  mean() - 3* std() ---下限
#  mean() + 3* std() ---上限

# 自实现3sigma 原则

def three_sigma(ser):
    """
    自实现3sigma 原则
    :param ser: 数据
    :return: 处理完成的数据
    """
    bool_id = ((ser.mean() - 3 * ser.std()) <= ser)  &   (ser <= (ser.mean() + 3 * ser.std()))

    # bool数组索引
    
    # ser[bool_id]

    return ser.index[bool_id]

#使用detail 验证
deatil = pd.read_excel("./meal_order_detail.xlsx")

print(deatil.shape)

# 调用3sigma原则，进行异常值过滤
index_name_list = three_sigma(deatil['amounts'])

deatil = deatil.loc[index_name_list,:]

print(deatil.shape)

#percentile() 计算分位数

# np.percentile() ql-1.5iqr  qu + 1.5iqr