Hello Chris, Great Project! Any Advice?

[6743152]

Hello Chris, Great Project! Any Advice?

[cbyn]

Hi William,

That was for loading a pre-trained model into the web app (see http://scikit-learn.org/stable/modules/model_persistence.html). The web app was just for promoting the project and didn't do anything particularly useful (it just displayed live streaming data with predictions). I don't know if I would expect anyone to get that running again.

The project is no longer maintained. I have only left it up as an example for people interested in similar research. Thanks for your interest though!

Chris

[cbyn]

I no longer have anything associated with the project other than what is in the repo. And if I did, it would be a model trained two years ago. If this is something that you want to pursue you will need to collect your own data and train your own model. There are many people who have forked the repo and have it in active development so you may want to look at what others are doing. This is not something that I am working on anymore. Good luck!

[cbyn]

I don't recall, but generating pkl files is standard python. cols.pkl is the list of columns used in the trained model: pred = model.predict(data[cols].values)[0].

[cbyn]

There is no push button way to do it. You will need to collect data for weeks, then train a model on that data (using the code in the repo). Once you have a trained model, pickling it (and the list of desired input columns) is trivial. Pickling the model for the web app is really the least of your concerns if you haven't even starting collecting data or training a model.