Aniket Sharma

Machine Learning Zoomcamp Update: Wednesday, 15 September 2021

Today, I completed the Sessions 2.6, 2.7, 2.8, 2.9 and 2.10. I completed it quickly as the topics were simple.

Session 2.6 - Linear regression: vector form

This session formulated the linear regression prediction equation in vector form.

Session 2.7 - Training linear regression: Normal equation

This session derived and explained the normal equation.

Key takeaways:

• Since, feature matrices are not square we cannot accurately calculate the weights to make predictions.
• We multiple the feature matrix with its transpose to make it square.
• Normal equation: W = (X^TX)^-1X^Ty

Session 2.8 - Baseline model for car price prediction project

This session created a baseline model.

Session 2.9 - Root mean squared error

This session defined the root mean squared error which is a metric for evaluating regression models.

Key takeaways:

• RMSE = √(¹⁄_m*∑_i=1^m (g(x_i) - y_i)²)

Session 2.10 - Computing RMSE on validation data

This session computed RMSE on validation data for the baseline model.

Estimated Time Taken: 35 minutes