In Multi-Regression Analysis , Discriminant Analysis and MT method ( Multi-Variable Analysis ) , there is a process to select effective variables.
(Some software does this process automatically. Some software does not have such function.)
This method is that to select the large number of the ratio of variance (F-value).
If there are many variances selecting by hand is difficult. Genetic Algorithm helps us for such case.
The method to use Design of Experiments to select variance is introduced in MT System (one of the methods of Quality Engineering )
It is using orthogonal array for two factors. "0" means that "to use the variance in the analysis." "1" means that "not to use the variance in the analysis."
"Experiment" in this way is that the combination of variance in the analysis.
After the "experiment", we find the effective variance.
Statistical Way of Making Hypothesis
Selection of Methods
NEXT Sparce ModelingTweet