Homogeneity analysis is a method that encompasses a general method for optimizing and finding linear sum model equations. Regression analysis, discriminant analysis, principal component analysis, canonical correlation analysis, and so on.
Homogeneity analysis is an old machine learning method because it deals with linear sums. The disadvantage is that it cannot be used for models that are better represented by complex curves.
As an advantage, since it is a simple model, the explainability and interpretability of AI is very high. The coefficients in the model can be used directly for explanation and interpretation.
In addition, since the model feels like hitting the data straight, outliers feel like they are out of the data, and are easy to find. This view is introduced in the pages Compressing High Dimensions to 2D with Regression Analysis Systems and Compressing High Dimensions to 2D with Canonical Correlation Analysis.
The term "homogeneity" is also used in addition to the above. I found the following on the net.
When I searched for "homogeneity analysis" on the Internet, I found quite a few explanations that "homogeneity analysis is the same as Correspondence analysis".
ANOVA (Analysis of variance) assumes that the variances for each group are considered equal.
The test to check whether this premise holds is Regarding the "Hypothesis Testing for Diffrence of Dispersion", there are materials in the world that call this "Test for Homogeneity".
NEXT Regression AnalysisTweet