Web-Based Fertilizer Recommendation System Using Soil, Crop, And Environmental Parameters With Random Forest Classifier

Abstract

The agricultural sector plays a crucial role in food security where crop productivity is highly dependent on the availability of nutrients in the soil such as Nitrogen (N), Phosphorus (P), and Potassium (K). However, inappropriate fertilization practices, whether insufficient or excessive, can damage soil structure and reduce crop yields. Farmers often have difficulty in precisely determining the most appropriate type of fertilizer based on specific soil and weather conditions. This study aims to build an intelligent recommendation system based on Machine Learning that can accurately predict fertilizer types. The algorithm used in this classification is the Random Forest Classifier which was chosen because of its ability to handle complex data sets and minimize overfitting. The software development method applied is the Agile method which allows for an iterative development process and is responsive to changing needs. The parameters used as input values include temperature, air humidity, soil moisture, soil color, plant type, and N, P, and K levels. The results of model testing using evaluation measurements show that the Random Forest algorithm is able to provide an accuracy level of 91.25% in predicting the right fertilizer class for various types of agricultural crops. This system can help farmers make more efficient and data-driven fertilization decisions