deepstack

keras implementation for spatio-temporal data.

The Project will house a Keras implementation of CNN, as well as, various 
geostatistical and spatial methods. 

Collaborators:

1. Megan Mason
2. Joel A Gongora
3. Monica Vermillion
4. Billy Henshaw
5. Michael Loso
6. Colin Quinn
7. Eric Morway
8. Ellen Van De Vijver
9. Fiona West

Objectives:

1. Gain knowledge in manipulating and managing multi-dimensional
   spatio-temporal datasets via python package.
2. Learn about the pipeline necessary to develop a
   deep learning product. 
3. 
3. Learn how to extract variable importance from a model

Scientific Questions:

1. What are the spatio-temporal patterns of accumulation and melt?
2. What is the relationship of different topographic variable to the 
   accumulation and melt patterns?
3. What is the relationsihp of different climatic variables to the 
   accumulation and melt in patterns?

Getting Started:

1. git clone https://github.com/geohackweek/ghw2019_deepstac.git
2. cd ghw2019_deepstac
3. run conda create -f environment.yml
4. bash ./scripts/download_s3_data.sh

utils

This folder contains scipts with modularized functions or classes. 

notebooks

This folder contains all complete notebooks.

1. jag This folder contains

models

Contains pickled models. 

data

This folder will contain the data you need for this project: 
however it
will be ignored by the .gitignore.