Comprehensive project focused on stock price prediction for Tesla stocks, leveraging advanced deep learning techniques. This project involved developing and comparing three different models: Long Short-Term Memory (LSTM), 1D Convolutional Neural Network (1D CNN), and Gated Recurrent Unit (GRU). The LSTM model achieved a high R2 score of 90%, demonstrating excellent predictive accuracy, while the GRU and 1D CNN models attained R2 scores of 72% and 63%, respectively.
To ensure optimal performance, I employed Keras Tuner for thorough hyperparameter tuning across all models. This process was critical in refining the models to achieve the best possible performance and robustness.
In addition to these models, I also experimented with a hybrid architecture combining LSTM and 1D CNN to further explore potential improvements in predictive accuracy.
For the implementation, I developed a Flask backend to handle the prediction logic, enabling efficient processing and serving of stock price predictions. Additionally, I built a React frontend to provide a seamless and user-friendly interface, allowing users to interact with the prediction models effortlessly.