- Developed a Bitcoin price forecasting project using TensorFlow, organising time series data into a supervised learning framework with precise windowing techniques.
- Evaluated multiple models (Naive Forecasting, Dense Models, LSTM, and Ensemble) with the ensemble model surpassing individual models, achieving an MAE of 565.04 versus a 567.99 baseline. Identified that outliers caused an extreme MAE of 17,134 in a 6th model.
Image credit: Designed by Freepik