[FIXED] How can I create columns out of a nested dictionary in python pandas

Issue

I'm trying to filter my data frame by making separate columns to increase readability and usability.

Problem statement: Column "Editables" has a nested dictionary as the value I'm trying to create separate columns as per the key. For instance 'photo_repace' 'text_remove', 'text_add' has different columns.

The nested dictionary:

{
'photo': {
    'photo_replace': None,
    'photo_remove': None,
    'photo_add': None,
    'photo_effect': None,
    'photo_brightness': None,
    'background_color': None,
    'photo_resize': None,
    'photo_rotate': None,
    'photo_mirror': None,
    'photo_layer_rearrange': None,
    'photo_move': None
},
'text': {
    'text_remove': None,
    'text_add': None,
    'text_edit': None,
    'font_select': None,
    'text_color': None,
    'text_style': None,
    'background_color': None,
    'text_align': None,
    'text_resize': None,
    'text_rotate': None,
    'text_move': None,
    'text_layer_rearrange': None
}
}

Output

Code I've been using:

df["editables"] = df["editables"].apply(lambda x : dict(eval(x)))
df_edit = df["editables"].apply(pd.Series )

Output:

Solution

Use json.json_normalize per rows and concat, last remove values before first . in columns names:

from pandas.io.json import json_normalize

c = ['photo.photo_replace', 'photo.photo_remove', 'text.text_remove']
df = pd.concat([json_normalize(x)[c] for x in df['editables']], ignore_index=True)
df.columns = df.columns.str.split('.').str[1]
print (df)
  photo_replace photo_remove text_remove
0          None         None        None
1          None         None        None

Answered By - jezrael