0
$\begingroup$

I am looking to use ridge regression to predict end of quarter sales revenue. My features are sales pipeline and revenue booked quarter to date. As the quarter progresses sales pipeline will naturally decrease and revenue booked quarter to date will increase. I’m trying to figure out if it’s best to build a model for each week or instead build one model and use it for the entire quarter. The goal is to predict the quarter end sales revenue using the current weeks features using the weekly model or a quarterly model. Thank you

$\endgroup$
1
  • $\begingroup$ How does ridge regression figure into this? $\endgroup$
    – Carl
    Commented Jan 22, 2021 at 2:50

1 Answer 1

0
$\begingroup$

You should build one model. The model should include either deterministic or random term(s) capturing the uniqueness of each week. Example: $$ {\rm Sales\ Revenue} = \beta_0 + \beta_1 * {\rm Time} + \beta_2 * {\rm Time}^2 + ... $$ The complexity of entertained models depends on the sample size.

Now, your interest in ridge regression signals your interest in applying regularization. Why, may I ask? Is the sample size so small?... If that is the situation, you should play with various forms of regularization (lasso, elastic nets, LAR, principal components regression) and choose the best one using something like cross-validation.

$\endgroup$
1
  • $\begingroup$ Thank you! Sample size will be roughly 1000 or so. I only have 4 features in the model. Scikit learn advised ridge regression cause I only have a few features and I know that they are all highly correlated to the label (what I am trying to predict). The other regularization options are a better fit for models where you have many more features and unsure that all are highly correlated (according to scikit learns documentation). I am really new to scikit learn and machine learning. Its been a blast learning how stats is the backbone to machine learning. Thank you! $\endgroup$
    – Jason
    Commented Jan 24, 2021 at 2:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.