Calculating mahalanobis distances in discriminant analysis

Question

I'm trying to calculate the mahalanobis distances for centroids plotted using discriminant analysis.  I have 20 groups (with 10 measurements for each group) and would like to calculate the mahalanobis distances between group centroids.

I've looked in jmp stats books and the only information that I can find is that Sqdist[0] is used in calculating mahalanobis distances, but there isn't any information on how to actually go about doing this.

Any assistance is much appreciated.

L

shahin · Answer

I don't think that is the correct answer, Laura, did you ever figure it out ? I am asking myself the same question.  The only thing that I could find were: 1) distances from individual points to the group centroid, and 2) values for the group means for individual covariates.  Nowhere does it output a multivariate mean, or a distance between the two multivariate means.

unknown · Answer

I don't know if this is what you need, but under the top triangle>score options, you can select save formulas.  This will give you columns (with formulas) for SqDist[0] which is the quadratic form needed in the Mahalanobis distance calculations and the Mahalobis distances from each centroid.  Page 505 of the stat and graph guide mentions these columns plus you will get a prediction column.  I hope this helps.