How to use index in pandas data frame?

Question

I use three columns from the data frame for indexing:

df = df.set_index(['col1','col2','col3'])

Then I try to get the row that corresponds to a particular index. I tried many different ways:

x = df.loc[(123,456,789)]

x = df.ix[(123,456,789)]

x = df.loc[123,456,789]

x = df.ix[123,456,789]

None of these options works. The error message that I get is:

raise e1
KeyError: 'MultiIndex lexsort depth 0, key was length 3'

I checked if the used value of index exists:

inds = []
for ind in df.index:
    inds.append(ind)

if (123,456,789) in inds:
    print 'in'
else:
    print 'out'

As the result I get in (so, this used value of index exists). What am I doing wrong?

Happy001 · Accepted Answer · 2014-07-11 14:47:07Z

2

After you sort the df, all your loc/ix methods should work.

df = df.set_index(['col1','col2','col3'])
df.sort(inplace=True)

answered Jul 11, 2014 at 12:10

Happy001

6,4032 gold badges26 silver badges17 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Should I sort the data frame before or after the indexing? How exactly should I sort it (what column should I use for sorting)?

After setting the index, sort the dataframe using index like above. Sorting will make MultiIndex lexsort depth (in the error message) from 0 to 3.