Chapter 8: Addressing Data Issues When Combining DataFrames

At some point during most data cleaning projects, the analyst will have to combine data from different data tables. This involves either appending data with the same structure to existing data rows or doing a merge to retrieve columns from a different data table. The former is sometimes referred to as combining data vertically, or concatenating, while the latter is referred to as combining data horizontally, or merging.

Merges can be categorized by the amount of duplication of merge-by column values. With one-to-one merges, merge-by column values appear once on each data table. One-to-many merges have unduplicated merge-by column values on one side of the merge and duplicated merge-by column values on the other side. Many-to-many merges have duplicated merge-by column values on both sides. Merging is further complicated by the fact that there is often no perfect correspondence between merge-by values on the data tables...