使用Numpy对特征中的异常值进行替换及条件替换方式

import numpy as np
 
data = np.array([['nan', 1, 2, 3, 4], # 数据类型为字符串型
        [10, 15, 20, 25, 'nan'],
        ['nan', 5, 8, 10, 20]])
print(data)
# [['nan' '1' '2' '3' '4']
# ['10' '15' '20' '25' 'nan']
# ['nan' '5' '8' '10' '20']]
 
data[data == 'nan'] = 100 # 将numpy中为'nan'的项替换为 100
print(data)
# [['100' '1' '2' '3' '4']
# ['10' '15' '20' '25' '100']
# ['100' '5' '8' '10' '20']]
 
data = data.astype(float) # 将数据由字符型转换为浮点型
print(data)
# [[100.  1.  2.  3.  4.]
# [ 10. 15. 20. 25. 100.]
# [100.  5.  8. 10. 20.]]

print(data)
# [[100.  1.  2.  3.  4.]
# [ 10. 15. 20. 25. 100.]
# [100.  5.  8. 10. 20.]]
 
data[:, 1][data[:, 1] < 5] = 5 # 对第2列小于 5 的替换为5
print(data)
# [[100.  5.  2.  3.  4.]
# [ 10. 15. 20. 25. 100.]
# [100.  5.  8. 10. 20.]]
 
data[:, 2][data[:, 2] > 15] = 10 # 对第3列大于 15 的替换为10
print(data)
# [[100.  5.  2.  3.  4.]
# [ 10. 15. 10. 25. 100.]
# [100.  5.  8. 10. 20.]]

import pandas as pd

data = {'hah':[1,2,9],
    '数量':[3,2,5],
    '价格':[10,9,8]}
df = pd.DataFrame(data)
df

import numpy as np
def panduan(x):
  x_mean = np.mean(x)
  print(x_mean)
  for i in x.index:
    if x[i] > x_mean*2:
      x[i] = x_mean
#     print(i)   
  return x

df = df.apply(lambda x:panduan(x),axis=1)