Community

Scikit Learn Machine Learning Tutorial For Investing - Pt. 19 - 98 Accuracy possible?

Hi Harrison,

I'm looking at part 19 of the aforementioned tutorial here:

https://www.youtube.com/watch?v=1dk9NO2M41E&index=19&list=PLQVvvaa0QuDd0flgGphKCej-9jp-QdzZ3

Using the NO_NA file, I'm getting a 98% accuracy, which seems odd given your circa-50% score. Can you guess as to what might be the problem, and why it is over-fitting?

I also notice this error:

UserWarning: Numerical issues were encountered when centering the data and might not be solved. Dataset may contain too large values. You may need to prescale your features.

I do, however, prescale the dataset using

preprocessing.sacle(X)

as you suggest.

Also, if you could elaborate a bit more on what exactly constitutes over-fitting in the context of this tutorial, I would appreciate it.

So, 3 questions in total:

1. What could cause a 98% accuracy? (supposedly, over-fitting)
2. What could cause the "Numerical Issues" error?
3. What exactly is over-fitting in this context?

Thanks,

-T