This message was deleted.
# general
s
This message was deleted.
k
Regarding hierarchical forecasting @Mário Amorim Lopes take a look to this [Probabilistic Hierarchical Forecasting with Deep Poisson Mixtures](https://arxiv.org/pdf/2110.13179.pdf). I am currently working on extending its baseline experiments on the TourismL and Favorita datasets.
m
That’s very interesting, thanks. We are using it to forecast pharmaceuticals based on the same molecule (and therefore, substitutes) instead of forecasting for each individual brand. That gives more leeway for drug stores to choose the vendor they want to promote in their commercial strategy