I don't know, although I have a very good experience in Regression analysis and Box-jenkins time series analysis. Best regards

Kindly check here...

https://www.quora.com/Why-does-Gaussian-discriminant-analysis-a-generative-learning-algorithm-have-%E2%80%9Cdiscriminant%E2%80%9D-in-its-name

Check here too..

https://stats.stackexchange.com/questions/4689/generative-vs-discriminative-models-in-bayesian-context

I really don't have any idea about, and thank you very much for the question. Tahar

Dear Professor,

Classification using generative models (e.g. naive Bayes, Fisher’s linear discriminant) is called discriminant analysis, because one is discriminating between classes or categories.

Classification using discriminative models (e.g. SVM or decision trees) is also sometimes called (non-parametric) discriminant analysis.

Although some methods such as LDA are not called discriminative, their generative models have been formed based on finding discriminative features. In other words, these methods use the parameters of the class models (a generative approach) to separate the classes (disriminative approach). This is why; Prof, Son said that other discriminative methods are called non-parametric whereas they may have some parameters e.g. regularization parameter in SVM

Discrimiant Analysis is a method for classification. Is using to evaluate the influence or ability that variables or atributes have to discriminante between class or categories with the assumptions that the sample space follows a multivariate normal distribution. You need indentify in your data set, if your outputs could be categorized. Regards

Classification problem