pandas: replace values in rows according to their amount-CodePudding

I have a dataframe as follows,

import pandas as pd

d ={'col1': [1, 2, 5], 'col2': [3, 4, 2], 'col3': [5, 1, 3]}
data = pd.DataFrame(d)

I would like to replace the max value with 'max', the min value with 'min' and the mid value (remaining value) with 'mid' in row-wise fashion. I have done the following:

data.apply(lambda row: row.replace(min(row),'min').replace(max(row), 'max').replace(r'\d ', 'mid', regex=True).astype('str'), axis=1)

while it works for min and max value, it does not work for mid value, but I am not sure what i am doing wrong.

CodePudding user response：

Use numpy.select for comapre max and min per columns and if values no match is assigned mid text:

arr = np.select([data.eq(data.max(axis=1), axis=0),
                 data.eq(data.min(axis=1), axis=0)], ['max', 'min'], 'mid')

df1 = pd.DataFrame(arr, index=data.index, columns=data.columns)
print (df1)
  col1 col2 col3
0  min  mid  max
1  mid  max  min
2  max  min  mid

Your code working if first convert to strings:

df1 = data.apply(lambda row: row.replace(min(row),'min').replace(max(row), 'max').astype('str').replace(r'\d ', 'mid', regex=True), axis=1)