Discriminant & TableOfReal: To TableOfReal (mahalanobis)...

Calculate Mahalanobis distances for the selected TableOfReal with respect to one group in the selected Discriminant object.

Settings

Group label
defines which group mean to use for the distance calculation.
Pool covariance matrices
when on, use a pooled covariance matrix instead of the group covariance matrix.

Algorithm

See Covariance & TableOfReal: To TableOfReal (mahalanobis)....

Example

Calculate the number of datapoints that are within the one-sigma elipses of two different groups, i.e. the number of data points that are in the overlapping area.

Suppose the group labels are ø and ʏ.

    pols50m = Create TableOfReal (Pols 1973): "no"
     Formula: ~ log10 (self)
     discriminant = To Discriminant
     selectObject: pols50m, discriminant
     t1 = To TableOfReal (mahalanobis): "\o/", "no"
     selectObject: pols50m, discriminant
     t2 = To TableOfReal (mahalanobis): "\yc", "no"

Now we count when both the t1 and t2 values are smaller than 1 (sigma):

    Copy: "tr"
     Formula: ~ object [t1] < 1 and object [t2] < 1
     Extract rows where column: 1, "equal to", 1
     no = Get number of rows"

© djmw, August 28, 2017