Time series forecasting with ARIMA and variants

Energy load forcasting project based on ARIMA statistical model and variants, developed for University exam

General info

This project use UKDALE dataset. You can visualize our predictions in results folder, there is 5 folders each on that rappresent one house and containts data in csv format and graphs. In particolar there are results with 7 days traing and 1 day prediction, 30 days traing and 1 day prediction for each household appliance.

In alternative you can run the code to get results, but in some case you might get different results because we have tuning ARIMA parameters for some household appliance to get better results.

How to use

First of all clone this repository git clone https://github.com/Vita98/EnergyLoadForecasting.git to make sure there are no missing folders

Arima

For arima you can use arima.py, arimaNotebook.ipynb or arimaAutomator.py. You can modify the configuration in the appropriate block in the code and in main block

Sarima

For sarima you can use sarima.py, sarimaAutomator.py or sarimaNotebook.ipynb. You can modify the configuration in the appropriate block in the code and in main block

Sarimax

For sarimax you can use sarimaxAutomator.py or sarimaxMulti.py. You can modify the configuration in the appropriate block in the code and in main block

Authors

• Alessio Tartarelli aka a-tartarelli
• Vitandrea Sorino aka Vita98