Preprocessing binary columns in pandas

Question

I am working with data frame in pandas.

import pandas as pd
df = pd.read_csv('Amazon_Historical_StockPrice2.csv',parse_dates=['Date'], index_col='Date')

I need to choose only binary columns for preprocessing. I was trying to make a condition for the value of the columns, to choose columns with values less or equal 1 and greater or equal 0, but this condition doesn't exclude rational numbers in this interval. I would like to ask, is there any option to make it automatically as there are too many columns to make it by hands?

Thanks!

Could you please give us an example of how to file is structure and example of a result you wish to get? — Lidor Eliyahu Shelef, Jan 09 '23 at 07:32

score 0 · Answer 1 · answered Jan 09 '23 at 07:46

import numpy as np
import pandas as pd
df = pd.DataFrame({
   "id": [100, 100, 101, 102, 103, 104, 105, 106],
   "A": [1, 2, 3, 4, 5, 2, np.nan, 5],
   "B": [45, 56, 48, 47, 62, 112, 54, 49],
   "C": [1.2, 1.4, 1.1, 1.8, np.nan, 1.4, 1.6, 1.5],
   "Binary": [1, 1, 0, 1, 1, 0, 1, 0]
})

df

    id    A    B    C  Binary
0  100  1.0   45  1.2       1
1  100  2.0   56  1.4       1
2  101  3.0   48  1.1       0
3  102  4.0   47  1.8       1
4  103  5.0   62  NaN       1
5  104  2.0  112  1.4       0
6  105  NaN   54  1.6       1
7  106  5.0   49  1.5       0

Use a list comprehension to get the column name of binary columns:

binary_cols = [col for col in df 
         if np.isin(df[col].dropna().unique(), [0, 1]).all()]

df[binary_cols]

Output:

Preprocessing binary columns in pandas

1 Answers1