Explore Outliers - Updated in JMP 16
See how to use the Explore Outliers utility, with emphasis on finding missing values and working with the Nines report and High Nines (missing) values. JMP 16 updates include:
- New Robust PCA Outliers utility (which replaces Multivariate Robust Outliers) is included in the Explore Outliers platform
- Multivariate Robust Outliers continue to be supported by JSL only and cannot be accessed through the menus.
- Robust PCA identifies cells, continues to respect the multivariate structure of the data, allows missing values, and is much faster than the Huber Mahalanobis feature used in previous JMP versions.
- Robust PCA supports creating a cleaned set of new columns, with imputed values for missing, and variously truncated values for the outliers.
Resources
1-5 minBasic JMPData PreparationUtilitiesVideoRe: Explore Outliers - Updated in JMP 16Updated link for Ledi's JMP On Air session: Exploring Outliers in Your Data
Re: Explore Outliers - Updated in JMP 16Thanks for the link share @wendytseng and for the intro video @ryandewitt.
Is there some rationale that we can point to for why the default values for Tail Quantile and Q, are 0.1, and 3, respectively?
From the JMP 16 Online Help Documentation: https://www.jmp.com/support/help/en/16.2/#page/jmp/quantile-range-outliers.shtml
Classically we know that values that fall beyond Q = 1.5 times the interquartile range past the Tail Quantile (or 1 minus the Tail Quantile) are identified as outliers, where the Tail Quantile is = 0.25 on the upper bound and 1 - 0.25 = 0.75 on the lower bound. This is the basis for the Outlier Box Plot:
