This repo contains my solution to Avertra's Forecasting Assignment. The solution is based on three notebooks:
- The first notebook contains some EDA and preprocessing.
- The second notebook is pretty large and exceeds the allowed file size in GitHub, so I uploaded it to Kaggle. The notebook presents the model development and performance evaluation stages.
- The third notebook displays the final results and performs the inference on new data.
In this work, different models are evaluated including:
- MLP
- LightGBM
- XGBoost
- CatBoost
- GRU
- LSTM
- Linear Regression
Surprisingly, the best performing model is the Linear Regression
model. This can be attributed to the simplicity of the task that renders complex models less effective.
Finally, insightful visualizations with some interpretations are presented.