Correlation and Outliers
Premium
Question: How would you measure correlation in the presence of outliers?
Outliers can distort correlation values by inflating or deflating the measured strength of the relationship. For instance, in the example provided, most x-values range between -7.5 and 5.0, and y-values between 7 and 13, with three outliers beyond that range. One approach is to use the Interquartile Range (IQR) method to detect and remove outliers. Standard normal scaling may also be applied, but is not ideal, as outliers can skew the mean and standard deviation, limiting the effectiveness of scaling.