Daniel Kostro

Can you give more details about what doesn't work, ideally a reproduction of the issue that we can test ourselves?

Hello! No iForest in not on our roadmap (to be honest we do not really have a roadmap, we just do things as we need them 😄) You're welcome to...

Cool! Would you like us to prepare a repo with some boilerplate on the mljs organization (with our up-to-date eslint rules, testing environment etc...), or would you prefer to start...

Hello @tommedema If your outcome is categorical then you want to look for "classification" algorithm, not "regression". Do you already know the outcome of you samples? If yes the keyword...

That's only if we are able to build a standalone wasm file. I'm not sure it's possible, because of file io methods that are called to serialize the model. https://github.com/emscripten-core/emscripten/wiki/WebAssembly-Standalone

Yes I think the dependencies should not be bumped in this one. There is an ongoing branch which updates some of the deps, so there will be conflicts https://github.com/mljs/libsvm/tree/upgrade-deps I'm...

Also the tests rely on code that is generated by emscripten, so no chance this will work without a build step first

The project is really outdated and would need to be reworked > 1. Which of the folder is your documentary demo page?? The demo page is generated in the demo-dist...

To make the build work, you'll have to use a unix system and install emscripten But you probably can just copy files that are published by hand if you don't...

Added with 0.2.0 Needs improved documentation and better error handling `predictInterval` should throw if not SVR or if model was not trained with probability estimates activated, idem for `predictProbability`