distribution_new_customer Enhancements
#1035
Labels
Comments
13 tasks
|
Hi @ColtAllen - I just made an attempt on this. Understand the it's low priority but it was easy enough to do this quickly. Have a look at my PR #1061 |
juanitorduz
added a commit
that referenced
this issue
Sep 23, 2024
* feat: test.txt added for commit check * feat: replaced plot_curve with plot_samples within ./mmm/plot.py * feat: n_samples added to distributions_new_customers * revert the plot.py changes --------- Co-authored-by: Juan Orduz <[email protected]>
jsnyde0
pushed a commit
to jsnyde0/pymc-marketing
that referenced
this issue
Sep 24, 2024
* feat: test.txt added for commit check * feat: replaced plot_curve with plot_samples within ./mmm/plot.py * feat: n_samples added to distributions_new_customers * revert the plot.py changes --------- Co-authored-by: Juan Orduz <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The
_distribution_new_customerinternal method was originally written for sampling from the latent purchase and dropout distributions ofCLVmodels. It is even more useful for posterior predictive checks, but is taking 1000 samples for each customer in the dataset, which is quite wasteful.This can be fixed by simply adding an
n_samplesparameter to_distribution_new_customer, and a default argument of 1 to thedistribution_new_customer_recency_frequencyexternal method. On that note,_distribution_new_customeris unique to each model, but the external methods that call it can probably be moved intoCLVBaseModel.The text was updated successfully, but these errors were encountered: