Implementation of linear regression, lasso/ridge regression, and kernel trick ridge regression in a real-life example

Linear regressions are simple and convenient tools that are applicable in modeling the relationship of multiple variables; this regression can also extrapolate or interpolate based on given values to predict other values. To investigate how linear regressions, linear regressions with kernel trick, and linear regression with lasso and ridge regularizations are applicable in various real-life situations, this paper will discuss the distinctions between the regressions and assess each of their efficacy by implementing them in a couple of multi-variate datasets, one involving crime rates and Boston housing attributes, the other concerned with fuel efficiency and car features. We conclude that linear regression performs the best in MSE(40.316) and RMSE(6.349) metrics, while the Lasso regression performs best in MAE(2.804) and MAPE(19.689) metrics.