sintel-dev/pyteller

Time series forecasting using MLPrimitives

pyteller's Issues

Store data using s3 bucket

It seems that the raw csv data files under this folder are pushed in the github.

I note that data.py this file actually supports downloading data from s3 bucket.

So the ideal solution could be removing all raw data files in all history commits. Otherwise, git clone will download a super large project.

Intermediate Outputs

Description

A feature should allow the user to specify which primitive they want to see an output from. This should be an option in the fit and forecast methods

What I Did

Previously, this was done with an extra training block in the output block in the pipeline json file. A better way to accomplish the task will follow GreenGaurd's approach of stepping through the pipeline