在数据框中循环遍历Pandas字典

Question

I have a dataframe with a column who's rows each contain a dict.

I would like to extract those dict's and turn them into dataframes so I can merge them together.

What's the best way to do this?

Something like:

for row in dataframe.column:
    dataframe_loop = pd.DataFrame(dataframe['column'].iloc(row), columns=['A','B'])
    dataframe_result = dataframe_result.append(dataframe_loop)

Answer 1

import pandas as pd

d = {'col': pd.Series([{'a':1}, {'b':2}, {'c':3}])}

df = pd.DataFrame(d)

>>>print(df)

      col 
 0  {'a': 1} 
 1  {'b': 2} 
 2  {'c': 3}

res = {}

for row in df.iterrows():
    res.update(row[1]['col'])

>>>print(res) 
{'b': 2, 'a': 1, 'c': 3}

Answer 2

If your column contains dicts and you want to make a dataframe out of those dicts, you can just convert the column to a list of dicts and make that into a dataframe directly:

pd.DataFrame(dataframe['column'].tolist())

The dictionary keys will become columns. If you want other behavior, you'll need to specify that.

Answer 3

I don't know what your dict in dataframe.column looks like. If it looks like the dictionary below, I think you can use pandas.concat to concentrate dictionaries together.

import pandas as pd

# create a dummy dataframe
dataframe = pd.DataFrame({'column':[{'A':[1,2,3], 'B':[4,5,6]}, \
                                    {'A':[7,8,9], 'B':[10,11,12]}, \
                                    {'A':[13,14,15], 'B':[16,17,18]}]})

#print(dataframe)

res = pd.concat([pd.DataFrame(row, columns=['A', 'B']) for row in dataframe.column], ignore_index=True)

print(res)