I am interested in Bayesian modeling but neither very experienced nor do I fully understand the different models and how they could be used.

I often have the use case to predict if it's likely that an event will happen based on some time series data from let's say the last couple of days.

An specific example would be predicting traffic jams based on traffic data collected the last month.

Another example would be to predict if it's likely that an owner of a house needs to turn on the lights in his living room based on his behaviour the last 3 days or so.

So the time series datas duration is rather short with only some samples. The model should be able to estimate the likelihood of the event (Traffic conditions: green/yellow/red, Light in Livingroom: on/off) happening. From what I've learned so far Bayesian models may be a good choice, but I find it quite difficult to find out which kind of models fit on that use case and which I should learn more about or if its feasible what I am trying to do.

I would appreciate if you could point me into the right direction, perhaps also aome hints of how to implement these models in software (I prefer to use python over R but I am able to use both).


One thing you could do is check out the bsts (Bayesian structural time series) library for R and its documentation. Facebook's prophet is also Bayesian and very powerful but slightly more complex.

I would suggest however to learn the basics of Bayesian inference before tackling a Bayesian time series method.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.