Thanks for the great work on pathpy! I do research on host intrusion detection with Linux system calls which can be nicely modeled with pathpy. However, a feature I am missing so far is the concept of uniform prior transition probabilities for higher order models.

In the current setup, when the likelihood of a path which contains a transition that has not been observed before is calculated, the likelihood will simply be zero. However, in my application it would be a lot more informative, if I had a uniform prior for all transitions. Because of the combinatorial explosion in higher order models, it is not possible to simply add all possible paths one time.

If there is interest from the community/maintainers, I could try to provide a PR for pathpy3 with this functionality. Any thoughts on this?