0
$\begingroup$

I want to predict how many number of days in the following week, an address will receive packages. So there are 8 labels, from 0 to 7. 0 means the address has no package. 7 means it gets packages everyday of the week. I don't have much information about the package and address. I only know the zip code for an address, the size and weight for an package.

Now I am building a multi-layer classification model with lag variable. I used past 10-week data for each address. Each week's data is a feature.

I also tried the address's geography, population density information as my label, but they don't have any help on the prediction.

I'm stuck on finding more useful features. Any suggestion on potential features and the model type itself?

$\endgroup$

1 Answer 1

0
$\begingroup$

Couple of features I could think of are distance between the distribution center(s) and the delivery address, demographics (gender, age, income, etc.), weather

$\endgroup$

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.