Please see my solution to the Predicting Bike Rentals guided project. I’d particularly appreciate your feedback on the conclusions I made after each model and at the end as I still find it a bit hard to interpret the metrics.
Predicting Bike Rentals.ipynb (226.4 KB)
Thanks in advance,
Click here to view the jupyter notebook file in a new tab
Congrats! Project looks good. Interpret the metrics is hard for all newbies, i still don’t feel that i can interpret them at 100%, but in my experience:
- Linear Regression: Works perfectly, both train and test error are similar. Obviously in MSE they are going to be really high because every value is squared, but RMSE have only a 0.4 point difference, so no problem at all
- Decision Tree: As every decision tree, it’s overfitted. But it is normal in DT. When min_samples_leaf = 5, the difference between train and test reduced but it’s still overfitted
- Random Tree: The first one is overfitted, the second one has a difference of 10 points i don’t consider it as an overfitted model but i’m not sure
A good explanation of overfit/underfit is found here
I’m agree with your final conclusion
That’s all i have to say, good luck!
Thanks, @alegiraldo666! I really appreciate your comments and the link you provided. I will make sure to go through it.
Have a lovely day!