I have a python dictionary in which the keys of the dictionary are tuples of two strings and the values are integers.

It looks like this:

mydic = { ('column1', 'index1'):33, 
          ('column1', 'index2'):34, 
          ('column2', 'index1'):35, 
          ('column2', 'index2'):36 }

The first string of the tuples should be used as the column-name in the dataframe and the second string in the tuple should be used as the index.

The dataframe from this should look like this:

Is there any way to do this?

(Or do I have to loop through all elements of the dictionary and build the dataframe one value at a time by hand?)

CodePudding user response：

Build a pd.Series first (which will have a MultiIndex), then use pd.Series.unstack to get the column names.

df = pd.Series(mydic).unstack(0)
print(df)

        column1  column2
index1       33       35
index2       34       36

CodePudding user response：

You can use pd.MultiIndex.from_tuples.

mydic = { ('column1', 'index1'):33, 
          ('column1', 'index2'):34, 
          ('column2', 'index1'):35, 
          ('column2', 'index2'):36 }

df = pd.DataFrame(mydic.values(), index = pd.MultiIndex.from_tuples(mydic))

                 0
column1 index1  33
        index2  34
column2 index1  35
        index2  36

What comes after that is just a workaround.

df.T.stack()

          column1  column2
0 index1       33       35
  index2       34       36

Notice that the index contains two rows. Do not forget to reset it.

df.T.stack().reset_index().drop('level_0', axis = 1)

  level_1  column1  column2
0  index1       33       35
1  index2       34       36

You can rename the level_1 if you want to. Hope it helps.