How to convert a pandas series of objects into a dataframe where each item becomes a column and the-CodePudding

I am stuck at this: I have a Series with the following structure Name: sn dtype: object

0 { key1: value1, key2: value2, key3: value3} 
1 { key1: value4, key2: value5, key3: value6} 
2 { key1: value7, key2: value8, key3: value9} 
3 { key1: value10, key2: value11, key3: value12}

I want to transform it into a DataFrame

   <key1>     <key2>    <key3>  # columns
0  <value1>   <value2>  <value3>
1  <value4>   <value5>  <value6>
2  <value7>   <value8>  <value9>
3  <value10>   <value11>  <value12>

I have tried using:

cols = ['key1','key2','key3']
df = pd.DataFrame(data=sn_sites['sn'], index=cols)

I was also able to iterate over the series and print item keys I need.

for index, value in sn.items():
  print (f{value['key1']}, {value['key3']) # works 
  df['column1'] = value['key1'] # doesn't work

But I still need a dataframe as output so I can progress on building my structure with columns from other dataframes and values like this inside series. Is there an simple way to do it?

CodePudding user response：

IIUC, this should work:

df = pd.DataFrame.from_records(sn)

CodePudding user response：

Your Series contains a dict in each row. Pandas can construct a DataFrame from structured input data like your Series using pd.DataFrame.from_records.

import pandas as pd

raw = [
    {"key1": 1, "key2": 2, "key3": 3},
    {"key1": 4, "key2": 5, "key3": 6},
    {"key1": 7, "key2": 8, "key3": 9},
]

# Each element is a dict
d_series = pd.Series(raw)

df = pd.DataFrame.from_records(d_series)

You can also unpack the Series right into the DataFrame constructor.

pd.DataFrame(*d_series)