Select rows in pandas MultiIndex DataFrame

Question

What are the most common pandas ways to select/filter rows of a dataframe whose index is a MultiIndex?

• Slicing based on a single value/label

Answer 1

Recently I came across a use case where I had a 3+ level multi-index dataframe in which I couldn't make any of the solutions above produce the results I was looking for. It's quite possible that the above solutions do of course work for my use case, and I tried several, however I was unable to get them to work with the time I had available.

I am far from expert, but I stumbled across a solution that was not listed in the comprehensive answers above. I offer no guarantee that the solutions are in any way optimal.

This is a different way to get a slightly different result to Question #6 above. (and likely other questions as well)

Specifically I was looking for:

1. A way to choose two+ values from one level of the index and a single value from another level of the index, and
2. A way to leave the index values from the previous operation in the dataframe output.

As a monkey wrench in the gears (however totally fixable):

3. The indexes were unnamed.

On the toy dataframe below:

    index = pd.MultiIndex.from_product([['a','b'],
                               ['stock1','stock2','stock3'],
                               ['price','volume','velocity']])

    df = pd.DataFrame([1,2,3,4,5,6,7,8,9,
                      10,11,12,13,14,15,16,17,18], 
                       index)

                        0
    a stock1 price      1
             volume     2
             velocity   3
      stock2 price      4
             volume     5
             velocity   6
      stock3 price      7
             volume     8
             velocity   9
    b stock1 price     10
             volume    11
             velocity  12
      stock2 price     13
             volume    14
             velocity  15
      stock3 price     16
             volume    17
             velocity  18

Using the below works, of course:

    df.xs(('stock1', 'velocity'), level=(1,2))

        0
    a   3
    b  12

But I wanted a different result, so my method to get that result was:

   df.iloc[df.index.isin(['stock1'], level=1) & 
           df.index.isin(['velocity'], level=2)] 

                        0
    a stock1 velocity   3
    b stock1 velocity  12

And if I wanted two+ values from one level and a single (or 2+) value from another level:

    df.iloc[df.index.isin(['stock1','stock3'], level=1) & 
            df.index.isin(['velocity'], level=2)] 

                        0
    a stock1 velocity   3
      stock3 velocity   9
    b stock1 velocity  12
      stock3 velocity  18

The above method is probably a bit clunky, however I found it filled my needs and as a bonus was easier for me to understand and read.