CODEWITHSUNDEEP
pythonpandasdataframemulti-index
user5421875
user5421875

Reputation:

Pandas Multiindex from array => TypeError: unhashable type: 'dict'

I'm trying to create the dataframe from the array with following structure:

df = [[{'date_time': Timestamp('2015-05-22 05:37:59'),
        'name': 'Tom',
        'value': '129'},
       {'date_time': Timestamp('2015-05-22 05:37:59'),
        'name': 'Kate',
        'value': '0'},
       {'date_time': Timestamp('2015-05-22 05:37:59'),
        'name': 'GroupeId',
        'value': '0'}, {...}, {...}, {...}],[another list of dictionaries like the first one],[and another one]]

using this code:

def create_from_arr():
    baby_array=pd.MultiIndex.from_tuples(df, names=['sessions', 'behaves'])
    return baby_array

I have the following error, that I couldn't understand:

TypeError: unhashable type: 'dict'

My desired output is like:

list 
                   date_time      name value
 1    0 2015-05-22 05:37:59       Tom   129
      1 2015-05-22 05:37:59      Kate     0
      2 2015-05-22 05:37:59  GroupeId     0
 2    3 2015-05-26 05:56:59     Hence   129
      4 2015-05-26 05:56:59      Kate     0
      5 2015-05-26 05:56:59     Julie     0
 3    ......................    ......  ......

Upvotes: 4

Views: 4910

Answers (2)

Anzel
Anzel

Reputation: 20563

I am still not sure what exactly you want to do with the MultiIndex, but here is one way to "flatten" your dictionary in your multi-level arrays and load your data into the dataframe properly:

Updated with "list" and "index" as MultiIndex

In [100]: data = [[{'date_time': Timestamp('2015-05-22 05:37:59'),
   .....:         'name': 'Tom',
   .....:         'value': '129'},
   .....:        {'date_time': Timestamp('2015-05-22 05:37:59'),
   .....:         'name': 'Kate',
   .....:         'value': '0'},
   .....:        {'date_time': Timestamp('2015-05-22 05:37:59'),
   .....:         'name': 'GroupeId',
   .....:         'value': '0'}], [{'date_time': Timestamp('2015-05-22 05:37:59'),
   .....:         'name': 'Tom',
   .....:         'value': '129'},
   .....:        {'date_time': Timestamp('2015-05-22 05:37:59'),
   .....:         'name': 'Kate',
   .....:         'value': '0'},
   .....:        {'date_time': Timestamp('2015-05-22 05:37:59'),
   .....:         'name': 'GroupeId',
   .....:         'value': '0'}]]

In [101]: df = pd.DataFrame(columns=['list', 'date_time', 'name', 'value'])

In [102]: for idx, each in enumerate(data, 1):
   .....:     temp = pd.DataFrame(each)
   .....:     temp['list'] = idx  # manually assign "list" index
   .....:     df = df.append(temp, ignore_index=True)
   .....:     
In [103]: df = df.reset_index()

In [104]: df.set_index(['list', 'index'])
Out[104]: 
                     date_time      name value
list index                                    
1    0     2015-05-22 05:37:59       Tom   129
     1     2015-05-22 05:37:59      Kate     0
     2     2015-05-22 05:37:59  GroupeId     0
2    3     2015-05-22 05:37:59       Tom   129
     4     2015-05-22 05:37:59      Kate     0
     5     2015-05-22 05:37:59  GroupeId     0

Upvotes: 3

Fabio Lamanna
Fabio Lamanna

Reputation: 21574

IIUC, let d be an extract of your array:

d = [[{'date_time': '2015-05-22 05:37:59',
   'name': 'Tom',
   'value': '129'},
  {'date_time': '2015-05-22 05:37:59',
   'name': 'Kate',
   'value': '0'}]]

I would extract the dataframe with:

df = pd.DataFrame.from_dict(d[0])

which returns:

             date_time  name value
0  2015-05-22 05:37:59   Tom   129
1  2015-05-22 05:37:59  Kate     0

Hope that helps.

Upvotes: 0

Related Questions