熊猫从字典中的矩阵创建数据框

Question

I have a dictionary Dict1 with keys as Dates and Sims.

Dates is an array with shape 100x1 and Sims has shape 100x5

I am trying:

import pandas as pd
df = pd.Dataframe.from_dict(Dict1) 

But errors out due to size of Sims. Is there a way I can create the DataFrame with each row of column Sims has size 5? ie each row can be stored as list or array of size 5.

Edit:

Dict1['Dates']
array([datetime.datetime(2016, 11, 1, 0, 0),
       datetime.datetime(2016, 11, 1, 1, 0),
       datetime.datetime(2016, 11, 1, 2, 0), ...,
       datetime.datetime(2025, 12, 31, 21, 0),
       datetime.datetime(2025, 12, 31, 22, 0),
       datetime.datetime(2025, 12, 31, 23, 0)], dtype=object)


Dict1['Sims']

array([[ 63.89694316,  35.8551162 ,  40.36134283, 57.23648392,
         35.96607425,  61.166471  ],
       [ 47.94894386,  53.95396849,  48.94336457, 51.04541849,
         28.69973176,  49.78683505],
       [ 63.90314179,  43.29467789,  36.97811714, 52.33639618,
         45.24190878,  69.9059308 ]...]])

Edit2:

I am looking to create the dataframe such that I can perform the following operation:

print(df[datetime.datetime(2016, 11, 1, 0, 0)])

[ 63.89694316,  35.8551162 ,  40.36134283, 57.23648392,
                 35.96607425,  61.166471  ]

Answer 1

You can use your Dict1['Dates'] as the index.

df = pd.DataFrame(Dict1['Sims'], index=Dict1['Dates'])
df.ix[datetime.datetime(2016, 11, 1, 0, 0)]

Note that you should use the df.ix[key] indexer, since df[key] defaults to looking up a column, not a row.

---

Alternatively, if you really want a single column containing list s, make sure that Dict1['Sims'] is a Python list, not a Numpy array before creating your data frame.

df = pd.DataFrame({'Sims': Dict1['Sims'].tolist()}, index=Dict1['Dates'])

The {'Sims': ...} construct tricks Pandas into interpreting the data as a single series of list s, rather than a multi-dimensional array.