This activity can stand alone and also is a companion activity to the Mastering JMP session Disentangling and Organizing Wide Data.
This hands-on activity allows you to practice using Explore Outliers to understand the extreme values in your data. The data are collected from historical data for predictive modeling purposes.
The instructions are below and in the attached PDF, which also contains the solutions. The Product Quality JMP data table is also attached to this post.
The data are in the Product Quality data table. They represent two response variables (Y1 and Y2) and nineteen predictor variables (X1 through X19) measured on 568 runs of a manufacturing process. You will examine the continuous predictor variables (X2 through X19) for potential univariate and multivariate outliers.
Answer the following questions:
- Are the potential univariate outliers in X2-X19 found with the default Quantile Range method and Robust Fit method the same?
- Are the potential multivariate outliers in X2-X19 found with the default Robust PCA method and K Nearest Neighbors method the same?
- Are any of the data values actual outliers?
Interested in a related activity on managing yourdata? See Explore Missing Values Hands-on Practice and Solution.