stoltzmaniac / basic-ml-oop Goto Github PK

When data is not normalized in LinearRegression.fit_stochastic_gradient_descent()

https://github.com/stoltzmaniac/Basic-ML-OOP/blob/master/01-Regression/Linear-Regression/Part-02/models/regression.py

The gradient blows up. Not only will it not converge, it will overflow. However, when normalized or standardized, it works just fine...

https://github.com/stoltzmaniac/Basic-ML-OOP/tree/master/01-Regression/Linear-Regression/Part-02

I have the models folder written up where data_handlers.py handles a bunch of the different inputs and have broken everything into classes. Then I have a separate dimensionalizers.py which does PCA. I also created regression.py to handle regression.

What I'm having trouble with is inheritance (OOP is not my strongest card). How can I improve this and utilize super() more efficiently? Currently, I cannot assign default values, I want the user to be able to input this at any point but have defaults set... I would also not like to force every single parameter to be input in the same order. I know this could be accomplished through *args and/or **kwargs but could really use a hand.