Transformation of trained model into an equation - Classification Learner

Problem Statement: Multi-class classification (3 classes)

Approach: Classification learner app - Ensemble (bagged trees) and SVM (cubic) are showing better results

Question: Using the parameters in the trained model, is there a way to develop an equation which can predict the classes? In other words, how to use the parameters in the trained model to make new predictions instead of directly using predict command? I need an equation to use it for control design. (Models are attached).

[Iuliu Ardelean]

Are you looking to actually write your own SVM and Ensemble models? Will the "equations" be used in MATLAB or somewhere else?

[Sadaf Batool]

I want to know the procedure that MATLAB adopts for classification using the parameters of the trained model. I want to know how these parameters are used for the prediction of the class.

Yes, I will be using the equation in MATLAB.

Answer 1

[Raynier Suresh]

One way is to take the equation from the documentation and replace the variable with the parameters from the trained model. You can refer the below links to know about the equation's

https://www.mathworks.com/help/stats/fitcecoc.html#bufogbr-5

https://www.mathworks.com/help/stats/ensemble-algorithms.html