ClimbsRocks / machineJS

[UNMAINTAINED] Automated machine learning- just give it a data file! Check out the production-ready version of this project at ClimbsRocks/auto_ml
https://github.com/ClimbsRocks/auto_ml
408 stars 64 forks source link

Persist everything to disk so a trained instance can be loaded #152

Open ClimbsRocks opened 8 years ago

ClimbsRocks commented 8 years ago

The goal is to train machineJS, serialize it and all of it's components to JSON/.pkl, then be able to load in all the trained components again on a different machine.

ClimbsRocks commented 8 years ago

Some things we'll have to save:

List of known steps required to write everything to disk and load up a fully-trained ppc instance at some future point:

data-formatter: min-max-scalers (one per column) data-formatter: imputing missing values what counts as a missing value what is it replaced by? this is just some code I picked off the internet, so we'll have to save and load this module by hand it shouldn't be hard though- just save one overall list of what is considered a missing value save an array, with one value for each column, that is the replacement value for that column. data-formatter: minimum date in date column