AN INTELLIGENT FARMLAND ADVISORY MODEL FOR BEST CROPPING AGRICULTURAL PRACTICES USING MACHINE LEARNING

Abstract

Agricultural practice is the major business activity of the well-being of people. Crop production is the important factor in agricultural practices, likewise soil pattern determines the suitability of crop to be cultivated. The cropping system in Nigeria faces challenges such as poor access to modern technology, climate change, inadequate digital infrastructure, and insufficient agricultural data for optimized crop management and productivity. This study aimed at developing an intelligent farmland advisory model for prediction of soil fertility for maize, guinea corn and millet crops using machine learning. The dataset used was 1550 data records. Dataset was pre-processed, cleaned, and divided into two; 80% for training while 20% for testing the model. Decision Tree technique and python Jupita notebook were used for classification and building the model. The model was evaluated using confusion matrix, precision, F1 score and recall. Precision and recall results were above 80% for both “Not fertile” and “Partially Fertile” in all the soil fertility tests for the three crops used. The performance of the model implies that it can be used to test soil fertility. The study recommended the implementation of the model to assist farmers in making informed decisions about crop cultivation based on soil suitability.