Machine learning-based land-use and land-cover classification using Google Earth Engine over a part of Haryana, India

It is the comparative study of three algorithms based on machine learning approaches (Random Forest (RF), Support Vector Machine (SVM) and Classification and Regression Tree (CART)) in Landsat 8 satellite imagery to classify Land Use and Land Cover (LULC) in a chosen area. The outcome shows that the Random Forest algorithm has the highest accuracy (87.5%) and good Kappa (0.777), and the results indicate that the algorithm has excellent classification results in all the classes of land cover and especially the Urban and Water classes. SVM showed competitive performance with more agreement (Kappa: 0.787) and unmatched accuracy exhibited in Urban regions whereas CART performed poorly with less accuracy (70.8%) and a major error of misclassification between Water and Vegetation classes. The results validate that Random Forest istruly the best and dependable classifier of LULC mapping in dissimilar landscapes