Model for Demand and Price Prediction of Agricultural Products in the Sacred Valley of Cusco Applying Machine Learning Algorithms

This study addresses the need for technological solutions in Peruvian agriculture, particularly in the Sacred Valley of Cusco, to optimize productivity and sustainability. A predictive model based on Machine Learning algorithms (Random Forest, XGBoost, Decision Tree and KNN) is proposed for the estimation of agricultural prices and demand. The data were obtained from the National Institute of Statistics and Informatics and preprocessed to ensure their quality. The performance of the algorithms was evaluated using four metrics: R2, MAE and MAPE. Random Forest proved to be the most efficient algorithm, with an R2 of 0.788 for “price” prediction and 0.937 for “demand”, excelling in accuracy and generalizability. The results suggest that the model can be a key tool for agricultural decision making, improving resource allocation and promoting sustainable practices. This work represents a step towards the integration of digital technologies in Peruvian agriculture, with positive implications for agricultural planning in diverse scenarios.