Enhancing product price prediction with mixed meta-heuristic algorithms for optimizing artificial intelligence models

Predicting agricultural product prices is crucial for farmers, distributors and policymakers to make informed decisions. Existing AI models for agriculture product price prediction struggle to account for the numerous factors affecting prices, such as weather conditions, market demand and seasonal variations. Predicting the price of agricultural products is difficult because of the limited availability of historical and real-time data as well as the intricate interaction of multiple unidentified factors that influence price. In this paper, we propose an association rule-driven light gradient boosting machine (AR-LGBM) to predict the price of the product. The agricultural crop pricing dataset collected from Virudhunagar District, Tamilnadu, India, was significant. Min-max normalization is used to pre-process the data. Principal Component Analysis (PCA) was utilized to extract features, which reduces data dimensionality while maintaining crucial data. Intelligent Flower Pollination Optimization (IFPO) is used for choosing the relevant features from the extracted features to simulate the proposed method using python. To evaluate the performance of the proposed method in terms of precision (87%), Mean Absolute Error (MAE) of 45.75, recall of 86%, Root Mean Squared Error (RMSE) of 2.15, F1-score of 85%, R-squared of 14, accuracy of 85% and Mean Squared Error (MSE) 32.72. As a result, the outcomes show that the suggested strategy outperforms traditional approaches. This study demonstrates the effectiveness of the AR-LGBM model in enhancing agriculture product price prediction.