This project develops a reverse prediction model to forecast 9 input variables from a single output, specifically focusing on data with periodic behavior. Utilizing advanced techniques such as sine and cosine transformations along with Fourier series for feature engineering, the model employs Random Forest, LSTM, and Conv1D algorithms to ensure accurate and robust predictions. This innovative approach aims to enhance understanding and forecasting capabilities in time series data.
- Reverse Prediction: Innovative approach to predict multiple input variables from a single output.
- Periodic Data Handling: Tailored to manage and forecast data exhibiting periodicity.
- Advanced Feature Engineering: Utilizes trigonometric and Fourier transformations to capture the underlying patterns in data.
- Multiple Algorithms: Leverages the strength of various algorithms including Random Forest, LSTM, and Conv1D for prediction.