The main idea behind this project comes from a post I read a few days ago:
"I think every data scientist should at least once spend some time coding the most widely used and fundamental ML models from scratch (regression, decision trees, random forest, boosting, NN’s). You will see a step change in your ability to work and solve problems with models if you do this and you will avoid a ton of pitfalls."
So that is exactly what we are going to do.
See full walkthrough here
See code implementation here
Slope change over time using Gradient Descent
See full walkthrough here
See code implementation here
Slope change over different lambdas