M.S. AAI Capstone Chronicles 2024

The analysis of these models suggests that while deep learning approaches like Transformer and LSTM offer sophisticated methods for time series forecasting, they may not always outperform simpler models like Linear Regression with seasonal data included, particularly when the dataset is small and univariate. Deep learning models often require larger datasets and more diverse features to uncover complex patterns effectively. In future work, expanding the dataset, exploring additional features, and experimenting with more advanced architectures could improve the performance of deep learning models. Additionally, incorporating multiple types of data and refining the training process may yield better results.

15

146