pymc-marketing
pymc-marketing copied to clipboard
Add `ExponentialGammaModel`
The Exponential-Gamma model is for continuous-contractual modeling of time-to-event scenarios; examples include the amount of time a customer is exposed to an ad campaign until conversion, and the amount of time a customer starts making in-app purchases after installing the app. A test dataset will need to be determined, and research literature is rather limited - all I was able to find is on page 5 of this paper:
http://mjbrouwer.pbworks.com/f/hfw_jof_98.pdf
However, implementation should be rather similar to the ShiftedBetaGeoModel. It is even mentioned a few times in the paper for that model:
https://faculty.wharton.upenn.edu/wp-content/uploads/2012/04/Fader_hardie_jim_07.pdf
Found a test dataset: https://www.brucehardie.com/datasets/hfw_trial_data.txt
Results from the above linked paper can be used for testing.
Just checking @ColtAllen is this the same paper: https://faculty.wharton.upenn.edu/wp-content/uploads/2012/05/Hfw_jof_98.pdf
the brouwer link above doesn't seem to work anymore
I think i'd be interesting in looking at this model if the issue is stale?
Hey @NathanielF,
That is the same paper; thanks for finding it. If you want to look into getting this added, that'd be great! The same model could probably even be expanded to support covariates via the approach described in this research note:
https://www.brucehardie.com/notes/019/time_invariant_covariates.pdf