I have finished today the project on KNN model applied to predicting car prices; it was an interesting one - the more I was testing, the more questions were coming out of the blue. One of the main struggles I had was related to the usage of
cross_val_predict, and how the two can yield very different results when assessing model accuracy, for instance with R2 as a metric.
The notebook can be seen here on my github profile.
I would be happy if you could share your opinion on the approach I have taken taken and if I got right the intricacies of cross_val_score and cross_val_predict. Early steps in machine learning give me plenty of doubts on whether I’m doing things right or straight up messing everything up