Extract values from dictionaries as records in dataframe columns-CodePudding

I have a dataframe which has several columns with values as dictionaries. These columns also contain some None (not sure if it's a string or just missing values).

df = pd.DataFrame([[{0: 300}, {0: 157}, {0: 456}],[{0: 298}, None, {0: 498}],[None, None, {0: 987}]],   columns=['col1', 'col2', 'col3'])

All dictionaries have key = 0, values = number from 100 to 10000.

I need to loop through all columns and rows and extract only the value. Preferably, i will overwrite the columns with only the values

So, end result should look like this:

df = pd.DataFrame([[300, 157, 456],[298, None, 498],[None, None, 987]],   columns=['col1', 'col2', 'col3'])

The number is actually an ID which I will use later on for a "vlookup" into another dataframe.

I tried with lambda functions:

df['col1'] = df['col1'].apply(lambda x: x.values() if x is not None else x)

I did manage to extract the values. Issue is the type of these values is recognized as a dictionary value (they look like this when i print them: (300)).

I need them as integers. I tried chaining an astype(int) but i do get an error (something like, you can't do that on a dictionary's value) Any thoughts?

CodePudding user response：

Use DataFrame.applymap for processing all columns with get first values of dict values:

df = df.applymap(lambda x: list(x.values())[0] if isinstance(x, dict) else x)
print (df)
    col1   col2  col3
0  300.0  157.0   456
1  298.0    NaN   498
2    NaN    NaN   987

If need integers convert values to Int64:

df = (df.applymap(lambda x: list(x.values())[0] if isinstance(x, dict) else x)
       .astype('Int64'))
print (df)
   col1  col2  col3
0   300   157   456
1   298  <NA>   498
2  <NA>  <NA>   987