Pandas: Convert DataFrame with MultiIndex to dict

pandas, python

Solution

df.stack().to_dict()

out：

{('DE', 'Lake', 'area'): 10,
 ('DE', 'Lake', 'count'): 7,
 ('DE', 'Forest', 'area'): 20,
 ('DE', 'Forest', 'count'): 5,
 ('FR', 'Lake', 'area'): 30,
 ('FR', 'Lake', 'count'): 2,
 ('FR', 'Forest', 'area'): 40,
 ('FR', 'Forest', 'count'): 3}

Problem

Another novice pandas question. I want to convert a DataFrame to a dictionary, but in a way different from what is offered by the `DataFrame.to_dict()` function. Explanation by example: ``` df = pd.DataFrame({'co':['DE','DE','FR','FR'], 'tp':['Lake','Forest','Lake','Forest'], 'area':[10,20,30,40], 'count':[7,5,2,3]}) df = df.set_index(['co','tp']) ``` Before: ``` area count co tp DE Lake 10 7 Forest 20 5 FR Lake 30 2 Forest 40 3 ``` After: ``` {('DE', 'Lake', 'area'): 10, ('DE', 'Lake', 'count'): 7, ('DE', 'Forest', 'area'): 20, ... ('FR', 'Forest', 'count'): 3 } ``` The dict keys should be tuples consisting of the index row + column title, while the dict values should be the individual DataFrame values. For the example above, I managed to find this expression: ``` after = {(r[0],r[1],c):df.ix[r,c] for c in df.columns for r in df.index} ``` How can I generalize this code to work for MultiIndices with N levels (instead of 2)? Answer Thanks to DSM's answer, I found that I actually just need to use tuple concatenation `r+(c,)` and my 2-dimensional loop above becomes N-dimensional: ``` after = {r + (c,): df.ix[r,c] for c in df.columns for r in df.index} ```

Original source