Session Ready
Exercise

Sensitivity to outliers

When we are analyzing the relationship of more than one variable, correlation is a great start. But how does correlation hold up against some more interesting datasets? How well does it hold up against outliers?

In this exercise, you will plot and compute the correlation for a dataset with an outlier and then remove it and see what changes. In the end, you want to see how correlation performs and come to a conclusion about when and where you should use it.

A sample dataset from the famous Anscombe's quartet has been imported for you as the df variable, along with the all the packages used previously in this chapter.

Instructions 1/3
undefined XP
  • 1
  • 2
  • 3
  • Display a matplotlib scatter plot of the X and Y features; notice the outlier in the top right.