列的最大值和最小值之间的差异

Question

I have a pandas dataframe with 2000+ columns. All the columns have numeric values. I want to find the difference between minimum and maximum values of each column. And then I want to filter top 10 columns having biggest differences.

Col1 Col2 Col3 ..... Col2500
 4     1    3  .....    6
 7     5   10  .....    17
 1    22    4  .....    2

I tried a few options, but none worked. Please suggest a solution.

Answer 1

This will give you the result in Series :

df.T.apply(lambda x: x.max() - x.min(), axis=1).nlargest(10)

Example:

df

   Col1  Col2  Col3  Col2500
0     4     1     3        6
1     7     5    10       17
2     1    22     4        2

df.T.apply(lambda x: x.max() - x.min(), axis=1).nlargest(3)

Col2       21
Col2500    15
Col3        7
dtype: int64

Or just:

(df.max() - df.min()).nlargest(10)

Answer 2

Here is my solution

>>> data = {'Col1':[4,7,1],'Col2':[1,5,22], 'Col3':[3,10,4], 'Col2500':[6,17,2]}
>>> df = pd.DataFrame(data)
>>> df
   Col1  Col2  Col3  Col2500
0     4     1     3        6
1     7     5    10       17
2     1    22     4        2
>>> diff = df.max() - df.min()
>>> diff
Col1        6
Col2       21
Col3        7
Col2500    15
>>> pd.DataFrame(diff).sort_values(by=0, ascending=False)
          0
Col2     21
Col2500  15
Col3      7
Col1      6

Answer 3

Hope this helps !

diff = df.max() - df.min()

diff.sort_values()

Example:

>>> df.values
array([[  0,  12,  42],
       [  1,  13,  21],
       [ 12,   1,  30],
       [  3,  45, -39],
       [  4,   1,  38]])

>>> diff = df.max() - df.min()
>>>
>>> diff.sort_values(ascending=False)
T3    81
T2    44
T1    12
dtype: int64
>>> diff.sort_values()
T1    12
T2    44
T3    81
dtype: int64
>>>

Answer 4

import pandas as pd
import numpy as np

#sample data
df = pd.DataFrame(np.random.randint(0,100,size=(100, 4)), columns=list('ABCD'))

#transposing data so columns are now rows and column names are indices
df = df.transpose()

#Calculation of Max - Min per row
df['dif'] = df.max(axis=1) - df.min(axis = 1)

#Number of results at the end (10 in your case)
TOP_N = 2

#Resetting the index to get column names and sorting by difference high to low
result = df.reset_index().rename(columns={'index':'ColumnName'})[['ColumnName','dif']].sort_values(by=['dif'],ascending=[False]).head(TOP_N)

print(result)