I have done all the feature engineering and am ready to start making a machine learning model that predicts the type of variable star based on its light curve. I have broken down this light curve into a series of points such as:
x, y (Time, brightness)
384, 12; 293, 43; 349, 12.
I'm hoping that the model will be able to classify the type of variable based on the shape of the light curve (which varies based on the type of variable star:)
But to do this, will I have to take a cross-product, or square the terms, or take the tangent of the terms? Or do I simply leave them as-is, and make a simple linear regression model? My guess is to cross-product, but I don't know if I should simply leave the cross-product term in there or add the time and brightness term for a total of 3 features.
