Colt Allen

Hey @Username159, `GammaGammaFitter.fit()` is a pre-fork legacy function from `lifetimes`; development ceased on that library 3 years ago. I no longer maintain `btyd` because I've transitioned these efforts to the...

[It seems the `summary_data_from_transaction_data` function is worth looking into as well](https://github.com/CamDavidsonPilon/lifetimes/issues/429). Any outstanding issues involving monetary value need to be resolved before `GammaGammaModel` is added.

After reviewing this paper, the monetary value calculations are being done properly: http://brucehardie.com/notes/022/RFM_summary_in_Excel.pdf However, a test should still be added to ensure the output of `calibration_and_holdout_data` matches that of the...

Hey @antocapp, There were some issues encountered with SciPy when testing on 3.10 a few months ago; possibly because the latest version was still a release candidate at the time....

Python 3.10 is still having `scipy` issues, but they seem to be isolated to the legacy `fitters.py` module: ``` Installing the current project: btyd (0.1b2) poetry run pytest ImportError while...

I'm planning to work on https://github.com/pymc-labs/pymc-marketing/issues/134 soon. The methodology for adding time-invariant covariates is rooted in survival analysis: https://www.brucehardie.com/notes/019/time_invariant_covariates.pdf

Hey @SSMK-wq, Customer churn is unobservable, so the only realistic validation approach is to set up a monitoring function to periodically review model predictions and verify if a customer has...

Thanks @SSMK-wq and @michaelwexler, I'll be updating the README.md file soon and also making an announcement in the `lifetimes` repo about this. Many of the improvements in `btyd` are being...

Hey @blackswanafrag, `lifetimes` was deprecated 4 years ago; `pymc-marketing` is the successor project: https://github.com/pymc-labs/pymc-marketing To answer your question, **non-repeat customers are only excluded when fitting a Gamma-Gamma Model.** The Pareto/NBD...

> Could you please develop a bit on the underlying statistical assumptions that says that the model should be fit on non-repeat customers as well? Every BTYD model is built...