multiindex

MultiIndex is a data structure in pandas that represents hierarchical indexing on an axis of a DataFrame or Series. It allows an axis to be addressed by a tuple of labels, with each label selected from a separate level. This enables sophisticated data organization where rows or columns can be grouped by multiple categories.

A MultiIndex comprises levels, codes, and optional names. Each level defines the set of labels for that tier, and codes are integer arrays that map each position along the axis to a label in the corresponding level. The combination of level labels forms the index values. Levels can have named identifiers to improve readability.

MultiIndex objects are typically created from lists of labels using constructors such as from_arrays, from_tuples, or from_product. They support operations including selection by tuples, stacking and unstacking to reshape data, and grouping by level via groupby. Other common operations include swapping or sorting levels, resetting to a flat index, and aligning indexes with reindex. These features enable multi-dimensional pivot-like analyses within a single axis.

Common use cases include time-series data with hierarchical categories (e.g., country and city), MultiIndex dataframes in which columns or rows carry multiple categorical dimensions, and complex aggregations. While powerful, MultiIndex can introduce complexity and memory overhead, and operations may be slower on large datasets compared to flat indexes. Careful design and occasional flattening can help balance expressiveness and performance.