This project was sponsored by Swanson School of Engineering. The objective is to integrate demand forecasting into inventory control policy making for medium-sized restaurants. An outcome of academic papar abstract was produced.
Literature contains the original codes and paper from reference studies Detailed content is mentioned in the abstract.
Project contains project code, some of code files were lost, and this study has not been finished because further development for model perfomance robustness is missing.
- Forecasting Model is for RNN algorithms forecasting (including LSTM, RNN, GRU) and the introduction of transformer in feature selection improved the performance by 2%.
- NonRNN is for other ML algorithms forecasting (including but not limited to: LR, Ridge Regression, SVR, SGD, DT, etc)
- Single Simulation is for debugging in simnulation logic because there was an explosion in result numerically due to issues in iteration, which was resolved later.
- Optimal Simulation is to obtain optimal parameters (reorder points) via searching algorithms search through space and get feedback from simulations.
- Proposed simulation is the unfinished script