Sustainable Smart Irrigation in Precision Farming using IoT

The current paper applies computational models based on machine intelligence to enhance the application of water in precision agriculture. This is to establish a reliable tool that can predict the most appropriate ambience of watering, hence enhancing sustainability. It includes what is referred to as sensor integration, and the establishment of a wireless sensor network (WSN). It gives analysis of a well-chosen data log containing essential sensor values like soil moisture, rain, humidity, wind speed and temperature. This paper work elaborates stages of Signal pre-processing, signal conditioning, training and testing performed for available stored dataset. Through the given 105,217 record dataset, smart irrigation accuracy compared using the support vector machine (SVM), random forest, decision tree, artificial neural network as well as other machine learning modes. Database are first trained then evaluated and finally samples are tested. The results of the current study depict that the random forest model is comparatively accurate than the other models with an accuracy of 92.68%. The second best performing method, which is only marginally less accurate than the random forest is decision Trees, with a success rate of 91.83%. SVM and neural network models also gave fairly good performance with accuracy of 90.78 and 90.21 respectively. The random forest model’s accuracy in forecasting ideal irrigation conditions which is further supported by the precision along with F1-Score measures. This paper provides a strong foundation for environmentally friendly water use and offers insightful information about using machine learning techniques for precision agriculture