Max depth diagrams look identical

M2. Selecting the best model M2. Overfit-generalization-underfit
#1

Hi,
in this section both diagrams at the end showing the Train/test errors look very similar, I cannot see where is the standard deviation for each curve.
Is it so small that it does not appear?
That’s why I don’t really understand how to use this .errorbar function.

Thanks

#2

In this case you can run the following line

scores[["train error", "test error"]].std()

The output would be

train error    4.979802e-15
test error     1.033944e+00
dtype: float64

In this case it is normal that the scale does not make the error bars really visible. In the notebook Effect of the sample size in cross-validation you will get some error bars that are easier to appreciate.

2 Likes