Set value multiindex Pandas

multi-index, pandas, python

Solution

First off, have a look at the docs on chained indexing

Second, read this about needing to sort MultiIndices.

That will get you to this solution:

In [46]: df = df.sort_index()

In [47]: df.loc['[21-23)', 'M', '[10000-20000)'] = 2

In [48]: df
Out[48]: 
                           A  B  C  D
Age     Sex Sumins                   
[21-23) F   [0-10000)      0  0  0  0
            [10000-20000)  0  0  0  0
        M   [0-10000)      0  0  0  0
            [10000-20000)  2  2  2  2
[23-25) F   [0-10000)      0  0  0  0
            [10000-20000)  0  0  0  0
        M   [0-10000)      0  0  0  0
            [10000-20000)  0  0  0  0
[25-27) F   [0-10000)      0  0  0  0
            [10000-20000)  0  0  0  0
        M   [0-10000)      0  0  0  0
            [10000-20000)  0  0  0  0
[27-29) F   [0-10000)      0  0  0  0
            [10000-20000)  0  0  0  0
        M   [0-10000)      0  0  0  0
            [10000-20000)  0  0  0  0

[16 rows x 4 columns]

pandas `.14` will have some additional ways for slicing a MultiIndex.

Problem

I'm a newbie to both Python and Pandas. I am trying to construct a dataframe, and then later populate it with values. I have constructed my dataframe ``` from pandas import * ageMin = 21 ageMax = 31 ageStep = 2 bins_sumins = [0, 10000, 20000] bins_age = list(range(ageMin, ageMax, ageStep)) indeks_sex = ['M', 'F'] indeks_age = ['[{0}-{1})'.format(bins_age[i-1], bins_age[i]) for i in range(1, len(bins_age))] indeks_sumins = ['[{0}-{1})'.format(bins_sumins[i-1], bins_sumins[i]) for i in range(1, len(bins_sumins))] indeks = MultiIndex.from_product([indeks_age, indeks_sex, indeks_sumins], names=['Age', 'Sex', 'Sumins']) cols = ['A', 'B', 'C', 'D'] df = DataFrame(data = 0, index = indeks, columns = cols) ``` So far all is well. I am able to assign value to a whole set of values ``` >>> df['A']['[21-23)']['M'] = 1 >>> df A B C D Age Sex Sumins [21-23) M [0-10000) 1 0 0 0 [10000-20000) 1 0 0 0 F [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 [23-25) M [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 F [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 [25-27) M [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 F [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 [27-29) M [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 F [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 ``` however, setting the value of one position only is a no go... ``` >>> df['B']['[21-23)']['M']['[10000-20000)'] = 2 >>> df A B C D Age Sex Sumins [21-23) M [0-10000) 1 0 0 0 [10000-20000) 1 0 0 0 F [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 [23-25) M [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 F [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 [25-27) M [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 F [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 [27-29) M [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 F [0-10000) 0 0 0 0 [10000-20000) 0 0 0 0 [16 rows x 4 columns] ``` What is going on here? I am open to the idea that i have completely misunderstood how multiindexing works. Anyone?

Original source