Problem Statement:
Implement Stochastic Gradient Descent for Linear regression MANUALLY with given SG formulas and apply it to Boston home price dataset; and then compare the results with actual Linear Regression from sklearn. This will give inner work of SGD. We can implement SGD for any algorithm once it is tried for Linear Regression.
Optimization problem without regulatization term and Derivatives or Gradient Descent.
Gradient is a slope. Instead of n in GD in diagram, we need to take bunch of k random points, suppose k=10, then we get SGD (Stochastic Gradient Descent) yi is actual value, wTxi is predicted value in below slope or derivative formulae.
