Outlier detection

Introduction

In this screencast we explore different criteria that can be used to flag outliers and we interpret those outliers in terms of probabilities.

learn what an outlier is and some common criteria for labeling points as outliers
understand the phenomenon of outlier masking and how to avoid it
learn to count and display upper and lower outliers according to various criteria

calculate outlier probabilities for (a) a single point and (b) at least one in $N = 260$ $N = 260$ points
- Note that I will not expect you to be able to do that last calculation but I do want you to appreciate that the probability of having at least one outlier in a sample is greater than the probability that any given data point will be an outlier
calculate the number of upper, lower, and total outliers using the $3\sigma$ criterion
display a boxplot of the measurement errors and compare the number of outliers to what you just calculated
calculate the number of upper, lower, and total outliers using the boxplot criterion
- Note that you can use the new BoxWhiskers function to help with this task. Check out the documentation.
calculate the number of upper, lower, and total outliers using the $3\, \text{MAD}_N$ criterion
display the rows that are outliers (any criterion)

After the screencast your spreadsheet should look something like the one below.