如果所有值都是某个字符串，则删除 pandas dataframe 中的列

Question

I have the following dataframe called df.

Sometimes it looks like this (where each value is unique)

   key   value1 value2 value3
0  key1  value  value  value
1  key2  value  value  value

Sometimes the entire column of value3 are is filled with dashes: '-' .

   key   value1 value2 value3
0  key1  value  value  -
1  key2  value  value  -

I want to find a command which will drop the column value3 if all the items in the column dashes.

I tried using df['value3'].any() and that that returns a '-' .

Is the right way to do what I want this?

if df['value3'].any() == '-':
    df = df.drop['value3'] 

Or is there a better way?

Answer 1

Use

>>> df.loc[:, ~df.eq('-').all()]
    key value1 value2
0  key1  value  value
1  key2  value  value

Answer 2

Compare for not equal by DataFrame.ne and then get all columns with match by DataFrame.any in boolean indexing :

print (df.loc[:, df.ne('-').any()])
    key value1 value2
0  key1  value  value
1  key2  value  value