Improved Predictive Model for Crop Recommendation and Crop Yield in Rural Areas using K-Nearest Neighbour (KNN) and Decision Tree (DT) Machine Learning Classification Algorithms

• Author(s): Joshua Chinemerem Chibueze; Agbakwuru Alphonsus Onyekachi
• Paper ID: 1713470
• Page: 1849-1864
• Published Date: 27-01-2026
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 9 Issue 7 January-2026

Abstract

The aim of this paper is to develop an improved predictive model for crop recommendation and crop yield in rural areas using K-Nearest Neighbour (KNN) and Decision Tree (DT) machine learning classification algorithms, to improve crop growth and yield through the analysis of the soil ph value as the target towards effective prediction and recommendation of crops using variables such as: crop types, rainfall, soil humidity and temperature are the major objectives of this paper. The study employed two machine learning classification algorithms namely: K-Nearest Neighbour (KNN) and Decision Tree (DT). The data was analyzed with JASP machine learning platform while the experiments are done using a dataset containing 2200 data?s sourced from Kaggle machine learning repository and was named as Hybrid_Agro_Crop_Recomender. During the experiment on the 2200 data?s, as contained in the dataset, 20% was split for test while 80% was split for train making up to a total of 100% after preprocessing was concluded. The result of the experiment showed a very high performance on both applied algorithms where (DT = 95% and KNN =94%). Decision Tree (DT) and K-Nearest Neighbors (KNN) produced F1 Score result for the following variables (apple 100%, banana 100%, blackgram 83%, chickpea 100%, coconut 97%, coffee 97%, cotton 97%, grapes 100%, jute 89% and kidneybeans 100%) accuracy for perfect soil compatibility with high percent nutrient to grow such recommended crops and on the result as produced by ROC curve, the accuracy shows that apple, banana and blackgram has a predictive accuracy from 80% to 100% while crops like maize, mango, mothbeans, mungbeans, mushmelon and orange has a predicted accuracy between 81% to 96%. From the result performance on both algorithms, the experiment shows that the use of a hybrid approach involving two or more classification algorithms in object classification is very much essential for effective decision making. Therefore the developed model was called Hybrid_Agro_Crop_Recomender as its application in decision making produced an excellent outcome to improve crop produce and health of the applied crops. Subject Area Artificial Intelligence, Intelligent Agriculture, Machine Learning, Decision Tree

Keywords

Artificial Intelligence, Machine Learning, Agriculture, Classification Model algorithms, and Crop recommendation model, Hybrid_Agro_Crop_Recomender, Crop Recommender Model

Citations

IRE Journals:
Joshua Chinemerem Chibueze, Agbakwuru Alphonsus Onyekachi "Improved Predictive Model for Crop Recommendation and Crop Yield in Rural Areas using K-Nearest Neighbour (KNN) and Decision Tree (DT) Machine Learning Classification Algorithms" Iconic Research And Engineering Journals Volume 9 Issue 7 2026 Page 1849-1864 https://doi.org/10.64388/IREV9I7-1713470

IEEE:
Joshua Chinemerem Chibueze, Agbakwuru Alphonsus Onyekachi "Improved Predictive Model for Crop Recommendation and Crop Yield in Rural Areas using K-Nearest Neighbour (KNN) and Decision Tree (DT) Machine Learning Classification Algorithms" Iconic Research And Engineering Journals, 9(7) https://doi.org/10.64388/IREV9I7-1713470