STUDY OF RESULT OF EXOPLANETS DATA GROUPING USING K-MEANS CLUSTERING AS A TRAINING FOR CLASSIFICATION

• Main Author: Harits Al Akbar, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/73498
• Institution: Institut Teknologi Bandung

Since the discovery in the early 1990s, exoplanets have been becoming a very interesting subject to discuss. Their existences seem to enlight a way to answer one of the most intriguing question: “are we alone in this universe?”. Until September 2022, there were more than 5000 exoplanets that have been confirmed by NASA. The data shows a wide diversity, and following a number of processes, the identification of characteristics for certain cluster of exoplanets can be revealed. In this research, analysis and grouping processes are carried out based on the exoplanet data on September 29, 2022. The data are then be processed with machine learning scheme by using unsupervised K-Means algorithm. The output are four clusters with their corresponding characteristics. This clustering process will later be used as the learning basis for a next implementation stage, namely a supervised learning using the K-Nearest Neighbors (KNN) algorithm. This classification stage will incorporate new data taken on May 11, 2023. Among results of the clustering process is a cluster that contain a group of exoplanets having some unusual physical properties. Moreover, result of the classification process using the K-Nearest Neighbors algorithm only shows that the new incorporated data is grouped into 3 clusters, apart from the odd one.