Create dictionary with pairs from column from pandas dataframe using regex

Question

I have the following dataframe

import pandas as pd
df = pd.DataFrame({'Original': [92,93,94,95,100,101,102],
             'Sub_90': [99,98,99,100,102,101,np.nan],
             'Sub_80': [99,98,99,100,102,np.nan,np.nan],
             'Gen_90': [99,98,99,100,102,101,101],
             'Gen_80': [99,98,99,100,102,101,100]})

I would like to create the following dictionary

{
    'Gen_90': 'Original',
    'Sub_90': 'Gen_90',
    'Gen_80': 'Original',
    'Sub_80': 'Gen_80',
 }

using regex (because at my original data I also have Gen_70, Gen_60, ... , Gen_10 and Sub_70, Sub_60, ... , Sub_10)

So I would like to create pairs of Sub and Gen for the same _number and also pairs or the Original with the Gens

How could I do that ?

Asked By: quant

||

Source

Answer 1

Use dictionary comprehension with replace and sorting by number after _:

d = {x:'Original' if x.startswith('Gen') else x.replace('Sub','Gen')
      for x in sorted(df.columns.drop('Original'), 
                      key=lambda x: (-int(x.split('_')[1]), x.split('_')[0]))}
print (d)

{'Gen_90': 'Original', 
 'Sub_90': 'Gen_90', 
 'Gen_80': 'Original', 
 'Sub_80': 'Gen_80'}

Answered By: jezrael

Answer 2

You can use:

cols = df.sort_index(axis=1).columns
group = cols[::-1].str.extract(r'_(d+)', expand=False)

out = {a: b for l in map(list, cols.groupby(group).values())
       for a,b in zip(l, ['Original']+l)}

This should work irrespective of the order of the input.

Output:

{'Gen_90': 'Original',
 'Sub_90': 'Gen_90',
 'Gen_80': 'Original',
 'Sub_80': 'Gen_80'}

Answered By: mozway

Answer 3

You can use a simple loop:

l = sorted(df.columns[1:])  # exclude 'Original' column as it's common
d = {}

# split other columns into two sublist: one for Gen_XX and another for Sub_XX
for g, s in zip(l[:len(l)//2], l[len(l)//2:]):
    d[g] = 'Original'
    d[s] = g

Output:

>>> d
{'Gen_80': 'Original',
 'Sub_80': 'Gen_80',
 'Gen_90': 'Original',
 'Sub_90': 'Gen_90'}

Answered By: Corralien

Answer 4

You can do:

gen_cols = df.filter(like='Gen_').columns
sub_cols = df.filter(like='Sub_').columns
d = dict(zip(sorted(sub_cols), sorted(gen_cols)))
d.update({g : 'Original' for g in gen_cols})
print(d)

{'Sub_80': 'Gen_80',
 'Sub_90': 'Gen_90',
 'Gen_90': 'Original',
 'Gen_80': 'Original'}

Answered By: SomeDude

Create dictionary with pairs from column from pandas dataframe using regex

Question:

Answers: