stampedpassp0rt / project-4 Goto Github PK

Hi Ameet -- Nice functional plotting code! Most students are not employing a functional coding style yet, so that definitely puts you in a good place. Also, really good explanations of your thought process regarding R-squared and why CV/Normalization gave such improvement.

One thing to take a look at: we can GridSearch l1 and l2 together, instead of running two models to look at just C values. You just include in the params_grid another key for 'penalty' and otherwise it works identically.

Re: how to evaluate coefficients and the model performance, since we’re using Logistic we might look at AUC ROC and Precision and Recall as opposed to accuracy score. There's a really great explanation here: http://mlwiki.org/index.php/ROC_Analysis

Nice work all around!