0
$\begingroup$

I'm looking to explain the revenue made by a coffee shop based on socio demographic data like gender, population, education and stores nearby. I've seen that by applying a logistic regression to revenue using those as explanatory variables could work but I don't finalize to understand how it is that I can give out number and compare each one of those variables to say which generates more revenue, could I apply a logistic regression to explain how much influence each one of variables selected have? And how? Also seen that Random Forest might be useful for it but I don't finish to get what it is and how can I apply it in this context?

$\endgroup$
0
$\begingroup$

A few things...

a) Why are you using logistic regression? Revenue is continuous and bounded below by 0, where as logistic regression is for binary outcomes.

b) It sounds like you are interested more by inference than prediction, so random forest -- though good at prediction -- really isn't what you want here.

My advice is to use something like a generalized linear model. I've seen Gamma regression used frequently to model financial problems, so that might be a first step.

$\endgroup$
  • $\begingroup$ a) Because I saw a couple of tutorials make it dichotomic, as saying more than 50 grand or less. But I actually want to keep it continuous. b) Yes indeed, I just want to know which demographic aspects influence more on final revenues for the shop. Gamma makes the best sense but glm library is not available for R-Studio 3.6, so it turns into a stack overflow question. Thanks! $\endgroup$ – Melania CB Jan 22 at 14:41
  • $\begingroup$ Would a single desition tree make sense also? $\endgroup$ – Melania CB Jan 22 at 14:52
  • $\begingroup$ Decision trees are nice because they are interpretable. You lose the ability to make confidence intervals though. $\endgroup$ – Demetri Pananos Jan 22 at 15:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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