Fuzzy Logic and Extreme Learning Machine for Web-Based Agribusiness Crop Recommendation

Abstract

Crop selection is not only an agronomic problem but also an operational decision in agribusiness because it affects input allocation, production risk, revenue potential, and supply-chain planning. This study develops a web-based crop recommendation system for agribusiness decision support by integrating Fuzzy Logic and Extreme Learning Machine (ELM). Fuzzy Triangular Membership Functions transformed seven crisp agro-environmental variables, namely N, P, K, temperature, humidity, pH, and rainfall, into 21 linguistic membership features. The transformed features were classified using an ELM model with 512 hidden neurons, sigmoid activation, and ridge-regularized analytic learning. The model was evaluated on the Crop Recommendation Dataset containing 2,200 records and 22 crop classes. The proposed Fuzzy-ELM model achieved an accuracy of 0.95, with macro-average precision, recall, and F1-score of 0.95, 0.95, and 0.94, respectively. The model was deployed in a Flask-based dashboard with MySQL storage and real-time weather integration to support practical crop-selection decisions. Black-box testing produced 100% valid results, while User Acceptance Testing reached 97.86%. The findings indicate that fuzzy feature transformation and fast analytic learning can support data-driven agribusiness applications under uncertain soil and climate conditions.