UK Imbalance Price Forecasting
This code is currently being reworked!
Purpose of this repo is educational - the code is focused on simple & understandable Python (using only functions, not classes).
Project is split into:
- elexton_data_scraping.py - pulls data from the Elexon API and saves in a sqlite database.
- make_dataset.py - pulls data from the sqlite databse and creates a machine learning dataset
- keras_models.py - contains functions to create feedforward & LSTM Keras models
- feedforward.py - creates a Keras model using the saved ff_data dataset
- lstm.py - creates a dataset & then a Keras model
Real data is included but it's unlikely you can train a very powerful model using only two features. This is meant as a demonstration of how to implement timeseries forecasting in Python.
Main dependencies for this project are keras (2.0.8) & tensorflow () or tensorflow-gpu (). Either CPU or GPU should run fine.
This project is built and maintained by Adam Green - adam.green@adgefficiency.com.