Patrik Schilter

As per Jeff's explanation, these variance estimates cannot be used for prediction intervals of CET, rather it requires simulating data for given parameters and predicting CET multiple times.

Besides the suggestion by @mmeierer to combine older cohorts into a single one there is also the more "hands-on" solution to manually set the estimation end to an arbitrary date....

See my issue discussing data structures as inputs to models #88

As discussed with Patrick, it might be more desirable to have entirely distinct classes for continuous- and discrete-time data. Reasons are to sensitize users for the differences and that the...

**Still existing issues** **Kink** Because of the dyncov model, the plotting points are currently set to be on the covariate dates. This often causes the second plotting point to be...

maybe also write a medium/towardsdatascience post that is a walkthrough with introduction to CLV, latent attrition models, etc

Postponed to next release. The kink is not so much of an issue and the shifted periods could not be found, yet.

Im not as familiar with the BG/NBD as I am with the Pareto/NBD but according to my knowledge, the heterogeneity in the transaction rates is also assumed to follow a...

> only 2 years worth of transactions One of the main strength of these models is that they normally require only very few transactions and periods for good results. Assuming...

You should be able to get confidence intervals using bootstrapping. Randomly selecting customers and all their respective transactions with replacement should be the way to go (and not randomly selecting...