Conference Proceedings
The International Conference on Emerging Technologies in Electronics, Computing and Communication 2022
(ICETECC`22)
Crops Recommendation with ANOVA-Based Feature Selection and Machine Learning Algorithms
Syed Ali Ishaq Mohsin1*; Syed Muhammad Ismail Hussain2; Hafiz Zia Ur Rehman1; Noman Naseer1;| 1Air University, Islamabad 2National University of Technology (NUTECH) |
ABSTRACT
Agriculture has been at the forefront in sustaining economies all over the world. However, challenges such as fluctuating climatic conditions, soil degradation, and inefficient crop selection practices continue to hold back productivity. Machine learning has been notably very successful in a large number of areas like healthcare, finance, and engineering; hence, it has immense potential to transform the sector into a data-driven agriculture one. This paper explains the application of ML models in optimizing crop recommendations based on some major agronomic and climatic factors such as nitrogen (N), phosphorus (P), potassium (K), pH, rainfall, temperature, and humidity. The study evaluates the performances of five popular machine learning algorithms: k-Nearest Neighbors (kNN), Gaussian Naïve Bayes (NB), Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM), using a dataset of 2,200 samples. Of these, Gaussian Naïve Bayes achieved the highest accuracy at 99.55%, thereby proving effective at handling feature-rich datasets. The Random Forest was also very satisfactory in performance; however, it showed important insight in feature importance. This study showed the critical role of exact feature selection in crop suitability and that ML can lead the drive in sustainable farming practices. Future work will incorporate external data sources, including satellite imagery and IoT devices, to increase the accuracy and adaptability of the models, resulting in more dynamic and holistic agricultural decision-support systems.