pandas allows the merging of pandas objects with database-like join operations using the pd.merge() function and the .merge() method of a DataFrame object. These joins are high performance and are performed in memory. A merge combines the data of two pandas objects by finding matching values in one or more columns or row indexes. It then returns a new object that represents a combination of the data from both based on relational-database-like join semantics applied to those values.

Merges are useful as they allow us to model a single DataFrame for each type of data (one of the rules of having tidy data) but to be able to relate data in different DataFrame objects using values existing in both sets of data.