At the beggining of this proyect I basically followed the path indicated in solutions. However there was a point, when making the task of normalizing the numeric columns, I went another way.
I did in the same way that showed previously in the mission course intead how it is made in this solution.
Somebody could tell me why in this solution it is used a diferent formula to normalize values in numeric columns?
Although I did this normalization in a different way, it seems the results are almost the same, so I think this is not as important.
Another question that for me is more important now is to really know what is the final objetive of all this proccess of selecting the best model for our predictions.
I mean, after selecting the best model: with its optimal features, k neighbour value, and adecuate k fold value validation. What would be the next steps if we want to do a true prediction for new data?
In this case for predicting prices for new cars for which we don’t know their actual prices. Should we use all our data as a train model for this prediction? And the bias and variability estimated would be those calcutated before for our choosen model?
Some help for solving this question would be very welcome.