R Tutorial: Outliers

Want to learn more? Take the full course at https://learn.datacamp.com/courses/correlation-and-regression-in-r at your own pace. More than a video, you'll learn hands-on coding & quickly apply skills to your daily work. --- Observations that don't seem to fit with the rest of points may be considered outliers. There isn't a universal, hard-and-fast definition of what constitutes an outlier, but they are often easy to spot in a scatter plot. In this scatter plot, we consider the relationship between the number of home runs hit by Major League baseball players in 2010 and the number of bases they stole. Home runs are a measure of batting power, while stolen bases are a measure of footspeed. It is not surprising that we see a negative relationship here, since power and speed are generally considered complementary skills. Since both variables here are integer-valued, several of the observations have the same coordinates, and thus the corresponding points are plotted on top of one another. This can misrepresent the data. To combat this, we can add an alpha transparency to the points, making them more translucent. Now we can see that the overplotting occurs where the darker dots are. Another approach is add some jitter to the plot. This is just a small amount of random noise in either the x or y direction. This relieves the constraint of having both coordinates be integers and thus allows us to see all of the data. In this plot, there are two points that stand out as potential outliers: the one in the lower-right hand corner and the one at the very top. We will discuss later in the course how to handle these outliers, but for now, it is enough to simply identify them and investigate them. In this case, we can use the filter function to identify those players with at least 60 stolen bases or at least 50 home runs. As it turns out, the player in the lower-right hand corner is Juan Pierre, who is one of the speediest and least powerful hitters in recent memory. The