Predictive modeling in Elixir/Nx?

Is anyone doing any kind of predictive modeling in Elixir based on real-world events?

Not sure how easy it would be but I have an idea for a site based around predicting the likelihood of a number of outcomes, hence wondered if anyone has any kind of experience they might like to share :blush:

Some examples for the sort of thing I had in mind:

  • Predict the likelihood of global warming passing a point of no return.
  • Predict the likelihood of another pandemic.
  • Predict the likelihood of nuclear war.
  • etc - sorry they’re so depressing! :lol:

It’s on my TODO list, mostly to forecast interest rates, FX rates, but more generally to forecast any signal with historical data - something like rainfall or temperature.

What I know so far - there are many techniques in common use: Monte Carlo, partial differential equations, lattice methods, Black Scholes, Ho Lee.

I suspect that ML models are out there. It seems like Nx/Scholar has many useful tools.

So - I’m also interested in predictive modeling and curious what others with more experience have to say on the subject. :wink:


What you want to do seems a lot saner/doable Andy! Please keep us posted :blush: