SMART IOT AND MACHINE LEARNING BASED APPROACH FOR CROP YIELD PREDICTION

Abstract

Machine learning approach provides rapid and efficient decision making on the basis of the data-driven databases. The real-time, continuous and precision approach using IoT based sensors can entails depth data recording without delay. The time-by-time crop status measurement provides the easy management for the crop critical factor in advance. This study provides a hybrid approach of the machine learning and IoT based data collection decision making for the crop yield prediction. The yield prediction by the driven datasets is obtained by the applying supervised machine learning models named SVM, Random Forest, Gradient Boosting and regression analysis. The comparative analysis is performed on the basis of accuracy, precision, recall and confusion matrix. The experiment is performed on the obtained dataset parameters includes weather (temperature, humidity, smoke and light intensity) and soil parameters (Soil pH level and Soil moisture). the season for the data collection is cotton crop cultivation time duration of the four months. the study shows the RF performed better result in context of accuracy, precision and recall. The comparative analysis shows RF can lead to a better prediction on basis of the collected dataset parameters values.