2023 Power Consumption Prediction AI Competition.
Developing an AI model that predicts the power consumption at a specific point in time using building information and spatiotemporal data.
- CPU: i7-11799K core 8
- RAM: 32GB
- GPU: NVIDIA GeForce RTX 3090 Ti
- Stratified Group KFold
- Ensemble Folds with Median
- Boosting is All you need
By default, hydra-core==1.1.0
was added to the requirements given by the competition.
For pytorch
, refer to the link at https://pytorch.org/get-started/previous-versions/ and reinstall it with the right version of pytorch
for your environment.
You can install a library where you can run the file by typing:
$ conda env create --file environment.yaml
Code execution for the new model is as follows:
Running the learning code shell.
$ sh scripts/run.sh
Examples are as follows.
python src/clustering.py
for model in xgboost lightgbm catboost; do
python src/train.py models=$model
python src/predict.py models=$model
done
python src/teach.py
python src/ensemble.py
XGBoost-custom-loss: 5.5316 LightGBM-tweedie-loss: 5.8699 Categorical-Non-Catboost: 5.5252 Categorical-Catboost: 5.7216
The NN model has a significant performance difference compared to the boosting. Ensemble results also appeared to have a greater impact than other models.
- meta feature: mean features
- forcasting model: NBeat is not performance