A Machine Learning Approach for the Land-Type Classification

Abstract

Land-type classification is an essential aspect of resource planning, and correctly classifying a land type can play a crucial role in designing and executing efficient utilization of land types for agriculture and other purposes. In recent years, machine learning (ML) methods have become popular, shown significant improvements in their performance and have been applied in various domains. In this work, we have proposed an ML-based approach for efficient and accurate classification of land types. We extensively experimented with different ML methods such as decision tree (DT), support vector machine (SVM), random forests (RF) and K-nearest neighbour (KNN). The empirical results suggest that ML-based approaches are superior for land-type classification. It is also found that out of the different ML methods applied on Statlog Landsat dataset, SVM outperforms other methods and achieves 92% accuracy better than other state-of-the-art methods.