In this repository one will find the files for our Non-Intrusive Load Monitoring Project in the Datascience Course at the University of Applied Sciences in Wedel. We tried three different approaches:
- LightGBM: An extensive library supporting the classification by gradient boosting methods.
- Keras: a Neural Net approach developed in Keras
- Neural Net: Another, second neural net
When tested, all approaches worked quite well. LightGBM was reasonable quick in training and detection, but the Keras implementation had shown a quite small improvement in regard to classification quality.