This repository contains the implementation of a multivariate time series forecasting model for predicting rice prices in Sri Lanka. The model incorporates various economic and non-economic factors and leverages the Temporal Fusion Transformer (TFT) algorithm to capture complex relationships between input features and rice prices. The project also includes thorough comparisons with other models such as SARIMAX and VAR. The insights generated by this model can assist in policy-making, supply chain management, and decision-making related to the Sri Lankan rice market.