This repository contains reproducable demo for our CSR and CNN models.

• numpy
• pandas
• scipy
• sklearn
• tensorflow
• tslearn

1. csr.py - Python implementation of our CSR model
2. cnn.py - Python implementation of our CNN model
3. train-forecasts.7z - Compressed set of demand forecasts used to train the models
4. test-forecasts.7z - Compressed set of demand forecasts used to test the models

1. First unzip the compressed data files
2. To run the csr model python csr.py this will execute with k = 300. You can pass -k argument to change the value of k.
   e.g. python csr.py -k 100
   We used euclidean distance for k-means clustering by default. You can change it to dtw by passing -dtw parameter
   e.g. python csr.py -k 100 -dtw
3. To run the cnn model python cnn.py