Python/Pandas Dataframe replace 0 with median value

问题

I have a python pandas dataframe with several columns and one column has 0 values. I want to replace the 0 values with the median or mean of this column.

data is my dataframe
artist_hotness is the column

mean_artist_hotness = data['artist_hotness'].dropna().mean()

if len(data.artist_hotness[ data.artist_hotness.isnull() ]) > 0:
data.artist_hotness.loc[ (data.artist_hotness.isnull()), 'artist_hotness'] = mean_artist_hotness

I tried this, but it is not working.

回答1:

I think you can use mask and add parameter skipna=True to mean instead dropna. Also need change condition to data.artist_hotness == 0 if need replace 0 values or data.artist_hotness.isnull() if need replace NaN values:

import pandas as pd
import numpy as np

data = pd.DataFrame({'artist_hotness': [0,1,5,np.nan]})
print (data)
   artist_hotness
0             0.0
1             1.0
2             5.0
3             NaN

mean_artist_hotness = data['artist_hotness'].mean(skipna=True)
print (mean_artist_hotness)
2.0

data['artist_hotness']=data.artist_hotness.mask(data.artist_hotness == 0,mean_artist_hotness)
print (data)
   artist_hotness
0             2.0
1             1.0
2             5.0
3             NaN

Alternatively use loc, but omit column name:

data.loc[data.artist_hotness == 0, 'artist_hotness'] = mean_artist_hotness
print (data)
   artist_hotness
0             2.0
1             1.0
2             5.0
3             NaN

data.artist_hotness.loc[data.artist_hotness == 0, 'artist_hotness'] = mean_artist_hotness
print (data)

IndexingError: (0 True 1 False 2 False 3 False Name: artist_hotness, dtype: bool, 'artist_hotness')

Another solution is DataFrame.replace with specifying columns:

data=data.replace({'artist_hotness': {0: mean_artist_hotness}}) 
print (data)
    aa  artist_hotness
0  0.0             2.0
1  1.0             1.0
2  5.0             5.0
3  NaN             NaN

Or if need replace all 0 values in all columns:

import pandas as pd
import numpy as np

data = pd.DataFrame({'artist_hotness': [0,1,5,np.nan], 'aa': [0,1,5,np.nan]})
print (data)
    aa  artist_hotness
0  0.0             0.0
1  1.0             1.0
2  5.0             5.0
3  NaN             NaN

mean_artist_hotness = data['artist_hotness'].mean(skipna=True)
print (mean_artist_hotness)
2.0

data=data.replace(0,mean_artist_hotness) 
print (data)
    aa  artist_hotness
0  2.0             2.0
1  1.0             1.0
2  5.0             5.0
3  NaN             NaN

If need replace NaN in all columns use DataFrame.fillna:

data=data.fillna(mean_artist_hotness) 
print (data)
    aa  artist_hotness
0  0.0             0.0
1  1.0             1.0
2  5.0             5.0
3  2.0             2.0

But if only in some columns use Series.fillna:

data['artist_hotness'] = data.artist_hotness.fillna(mean_artist_hotness) 
print (data)
    aa  artist_hotness
0  0.0             0.0
1  1.0             1.0
2  5.0             5.0
3  NaN             2.0

回答2:

use pandas replace method:

df = pd.DataFrame({'a': [1,2,3,4,0,0,0,0], 'b': [2,3,4,6,0,5,3,8]}) 

df 
   a  b
0  1  2
1  2  3
2  3  4
3  4  6
4  0  0
5  0  5
6  0  3
7  0  8

df['a']=df['a'].replace(0,df['a'].mean())

df
   a  b
0  1  2
1  2  3
2  3  4
3  4  6
4  1  0
5  1  5
6  1  3
7  1  8

回答3:

data['artist_hotness'] = data['artist_hotness'].map( lambda x : data.artist_hotness.mean() if x == 0 else x)

回答4:

Found these very useful, although mask is really slow (not sure why).

I did this:

df.loc[ df['artist_hotness'] == 0 | np.isnan(df['artist_hotness']), 'artist_hotness' ] = df['artist_hotness'].median()

来源：https://stackoverflow.com/questions/37506488/python-pandas-dataframe-replace-0-with-median-value

标签

python

pandas

dataframe

mean

median