After the calculation of Mahalanobis' Distance of MT method , we need the next step for Prediction of Abnormal.
In this approach, if more than the max of MD of Unit Space , it is abnormal.
In my experience, real data is not beautiful ditribution. So this approach is enough.
Sample file has the function of this approach.
If "more than Max(MD^2 of unit space)", decision is "C".
I put "A" and "B".
It is decided more than 1 or not.
"1" is the average of MD^2 of unit space.
If ration of "B" is less, unit space may include abnormal value.
NEXT 
Analysis of the cause of abnormalities by MT method
