Coeff for loop code

M4. Linear Models M4. Regularization of linear regression model
#1

coefs = [est[-1].coef_ for est in cv_results[“estimator”]]
what is est[-1].coef before the for loop ? i’ve never seen somthing before a for loop

1 Like
#2

I assume that est is a scikit-learn Pipeline made of a transformer followed by the linear model.
est[-1] extract the fitted linear model and access to its coefficient.

#3

Just adding that I found this very confusing too. It took me a while to figure out that what is happening is the following:

  1. Get list of pipelines from cv_results
  2. Loop over pipelines
  3. For each pipeline (it’s not an estimator, despite what the variable name implies!), retrieve the last item, corresponding to the estimator: the linear model (the other item is the polynomial features transformer)
  4. Get the array of coefficients
  5. The list comprehension stacks them all together, and later we stuff this into a DataFrame

It wasn’t trivial for me to unpack this, and I found it to be a significant distraction from the learning objectives for this notebook. It might be helpful to add some kind of explanatory note in the notebooks for future iterations of this MOOC.