The nsmallest and nlargest functions are extremely useful for returning the n smallest and n largest rows after ordering by the desired column.

In the sample sales data, let's find the 3 smallest records by Profit:

sales_data.nsmallest(3, "Profit")

The following will be the output:

The ordering can also be done with respect to multiple columns, as shown here:

sales_data.nlargest(3, ["Quantity", "Profit"])

The following will be the output:

These functions – nsmallest and nlargest have the keep parameter to decide how duplicates are handled. It helps to choose either the first occurrence, the last occurrence, or to retain all duplicates.