CODEWITHSUNDEEP
python-3.xpandasdataframe
tktktk0711
tktktk0711

Reputation: 1694

python3: delete the columns that the value of columns is a certain value

The is a dataframe(df1) like the following:

 eye_l   eye_r    nose       mouse       ear     eyebrow_l  eyebrow_r  ...
  34_35_a -1_-1_-1  45_66_b    45_64_a     78_87_a  -1_-1_-1   -1_-1_-1
  35_38_a -1_-1_-1  75_76_b    -1_-1_-1     38_79_a -1_-1_-1 -1_-1_-1
  64_43_a -1_-1_-1  85_66_b    65_45_a     87_45_a  -1_-1_-1  -1_-1_-1
   ....................................................................

I want to just delete the columns that all the value of a column is -1_-1_-1 (such as eye_r, eyebrow_1, eyebrow_r), and please notice that some columns may have a number (not all) of value is -1_-1_-1 will be kept.

I know that there is code like:

df1. drop(['eye_r', 'eyebrow_l', 'eyebrow_r '], axis=1, inplace=True)

But all the value is -1_-1_-1 of columns is not just three, there may be 100, and also there is a number of dataframes like such kind of dataframe. I want to process this issue in common method. Thanks

Upvotes: 1

Views: 36

Answers (1)

jezrael
jezrael

Reputation: 863501

Use boolean indexing with DataFrame.all and change condition to != or DataFrame.any with inverting mask by ~:

df = df.loc[:, (df != '-1_-1_-1').any()]

Or:

df = df.loc[:, ~(df == '-1_-1_-1').all()]

print (df)
     eye_l     nose     mouse      ear
0  34_35_a  45_66_b   45_64_a  78_87_a
1  35_38_a  75_76_b  -1_-1_-1  38_79_a
2  64_43_a  85_66_b   65_45_a  87_45_a

Detail:

print (df != '-1_-1_-1')
   eye_l  eye_r  nose  mouse   ear  eyebrow_l  eyebrow_r
0   True  False  True   True  True      False      False
1   True  False  True  False  True      False      False
2   True  False  True   True  True      False      False

print ((df != '-1_-1_-1').any())
eye_l         True
eye_r        False
nose          True
mouse         True
ear           True
eyebrow_l    False
eyebrow_r    False
dtype: bool

Upvotes: 3

Related Questions