How to convert a dictionary into a pandas dataframe with key and values in two separate columns?-CodePudding

I have a python dictionary in this format.

d = {1: {1, 2, 3},
     2: {4, 5}}

I want to convert it into a pandas dataframe in this format.

Expected Output:
    Source      Target
    1           1
    1           2
    1           3
    2           4
    2           5

I have tried doing this using list comprehension

d = {1: {1, 2, 3}, 2: {4, 5}}

df=pd.DataFrame([[key,v] for key, value in d.items() for v in value], columns=["Source", "Target"])
print(df)

But, is there any better way of doing this?

CodePudding user response：

You can use df.explode:

import pandas as pd

d = {
    1: {1, 2, 3},
    2: {4, 5}
}

df = pd.DataFrame(d.items(), columns=['Source', 'Target'])
df = df.explode('Target')

Which gives

   Source Target
0       1      1
0       1      2
0       1      3
1       2      4
1       2      5

Here, we create the dataframe with multiple values for each Target, and explode then creates a new row for each value in target.

Notice that the index still reflects the original dataframe, so we can use:

df = df.reset_index(drop=True)

To reset it to

   Source Target
0       1      1
1       1      2
2       1      3
3       2      4
4       2      5

Which combined gives us

df = df.explode('Target').reset_index(drop=True)

CodePudding user response：

You could create the DataFrame from each key:value pair in the dictionary and then concat them together.

import pandas as pd

pd.concat([pd.DataFrame({'Source': k, 'Target': tuple(v)}) for k,v in d.items()],
          ignore_index=True)

Or, you can use the pd.DataFrame.from_dict constructor, and stack, with a bunch of renaming

(pd.DataFrame.from_dict(d, orient='index')
   .stack()
   .reset_index(-1, drop=True)
   .rename('Target').rename_axis(index='Source')
   .reset_index()
   .astype(int))

   Source  Target
0       1       1
1       1       2
2       1       3
3       2       4
4       2       5