This beginner-level workshop will guide you through implementing a sales forecasting neural network model based on an openly available dataset. We will go through some basic neural network theory and apply a systematic model improvement approach to get from a simple model to a more complex one, trying out various tricks along the way.

Requirements

- python: beginner level
- a laptop with some software configured (see setup instructions)
- pandas, numpy, sklearn: nice-to-have

Setup instructions

You will need:
- a clone of: https://github.com/gregaw/sales-forecasting-with-nn.git
- anaconda (https://conda.io/docs/user-guide/install/index.html)
- [optional] google cloud ml-engine (there's free tier available) and shell (https://cloud.google.com/sdk/)

Once you get those installed (and have this git repo cloned locally), run:
> conda create -n mlengine python=2.7 anaconda
> source activate mlengine
> pip install -r requirements.txt

Hosted by Grzegorz Gawron