Comparison of the Naïve Bayes and the Nearest Neighbor method to detect vegetation in some of the New Capital of Indonesia (IKN)

Human activities or anthropogenic activities can cause land cover changes in an area. IKN as the new capital city of Indonesia has the potential to experience changes in land cover due to massive infrastructure development in the future. Land cover detection using remote sensing and machine learning is growing rapidly because it can monitor environmental conditions in an area. Planet imagery is one of the most recent remote sensing data with sufficient spatial resolution to identify land cover. This research aims to detect vegetation objects and built-up land in the study area. The data used is Planet imagery and the research area is part of the IKN in Kalimantan, Indonesia. The method proposed in this study is the object-based classification (OBIA) Naïve Bayes compared to Nearest Neighbor method. The results show that OBIA using these two machine learnings are able to detect vegetation and built-up land where the overall accuracy with Nearest Neighbor reaches 87.8% and Naïve Bayes obtains a higher accuracy of 98.7%. This research only classifies vegetation and built-up land and in the future research can be developed to see how biodiversity conditions in the IKN area can be detected. These results indicate that the proposed method is effective for detection of built-up land cover and vegetation as a solution for environmental monitoring in IKN. By considering Strengths, Weaknesses, Opportunities, and Threats and comparing these two methods, it is concluded that Naïve Bayes is better than Nearest Neighbor to detect vegetation in study area.