When using the agg method to perform an aggregation on multiple columns, pandas creates an index object with two levels. The aggregating columns become the top level and the aggregating functions become the bottom level. Pandas displays MultiIndex levels differently than single-level columns. Except for the innermost levels, repeated index values do not get displayed on the screen. You can inspect the DataFrame from step 1 to verify this. For instance, the DIST column shows up only once but it refers to both of the first two columns.

Step 2 defines new columns by first retrieving the underlying values of each of the levels with the MultiIndex method get_level_values. This method accepts an integer identifying the index level. They are numbered beginning with zero from the top/left. Indexes support vectorized operations, so we concatenate both levels together with a separating underscore. We assign these new values to the columns attribute.

In step 3, we make both index levels as columns with reset_index. We could have concatenated the levels together like we did in step 2, but it makes more sense to keep them as separate columns.