Optimizing Irrigation with AI: A Sensor-Driven Machine Learning Framework for Sustainable Agriculture

An efficient use of water is crucial in farming sustainability, in areas that have limited water. Fixed schedules or manual judgment, which is the basis of the traditional irrigation systems, usually lead to an over-or under-irrigation that influences the yields and waste resources. This paper suggests a smart irrigation system that can be performed through a machine learning system with the help of sensor data to make ideal irrigation choices. Six monitored algorithms, logistic regression, random forest, gradient boosting, AdaBoost, support vector machine, K-nearest neighbors machine learning related 2000 samples (irrigation_machine.csv) measured over three agricultural plots with 20 sensor characteristics and irrigation answers of either 0 or 1. Normalization, one-hot encodings and train-test split 70-30 were employed in preprocessing. The accuracy, precision, recall, and F1-score were used to measure performance. The best results were achieved by AdaBoost where 99.45 was recorded as the accuracy, 99.45 was the precision, 99.47 was the recall, and 99.52 was the F1score. The ability to get feature relevance and class balance insight by using correlation heatmaps was observed. This could be done in addition to making accuracy and efficiency more enhanced although there was no weather data. The suggested AI-based system effectively provides a high level of predictive performance, which minimizes the amount of unused water and promotes efficient precision irrigation at any scale and in any agricultural environment.