Implementation of Deep-Forecast using PyTorch

• Deep Forecast: Deep Learning-based Spatio-Temporal Forecasting
• Adapted from original implementation

Setup

• Clone this repository : git clone https://github.com/Wizaron/deep-forecast-pytorch.git
• Download and install Anaconda or Miniconda
• Go to the "reseg-pytorch/code/pytorch" : cd reseg-pytorch/code/pytorch
• Create environment : conda env create -f conda_environment.yml
• Activate environment : source activate deep-forecast-pytorch

Code Structure

• "data" : Stores data and scripts to prepare dataset for training.
• "lib" : Stores miscellaneous scripts for training and testing.
  • "arch.py" : Defines network architecture
  • "model.py" : Defines model (Minibatching mechanism, optimization, criterion, fit, predict, etc.)
  • "prediction.py" : Metrics and plots to evaluate the performance of the trained model
  • "data.py" : Creates training, validation and testings datasets
  • "loader.py" : Creates Dataset loader for PyTorch
• "train.py" : Main training script.
• "test.py" : Main testing script.
• "settings.py" : Defines hyper-parameters of the model.

Data

• Data is downloaded from IEM
• Download data and save it under "data/raw"
• To prepare dataset, run the scripts in "data/scripts"

Training and Testing

• Train : python train.py --data [PATH OF PREPARED DATASET]
• Test : python test.py --data [PATH OF PREPARED DATASET] --model [PATH OF THE SAVED MODEL]
• For more info : python train.py --help, python test.py --help

train.py

• It saves models and logs under "models"
• At the end of the training, it saves predictions under "outputs"

test.py

• It saves predictions under the directory of the model.

Sample Outputs