KOMPARASI ALGORITMA RANDOM FOREST DAN NAIVE BAYES PADA INDOOR POSITIONING SYSTEM DENGAN METODE FINGERPRINTING (STUDI KASUS : GEDUNG KULIAH TERPADU UIN SUNAN KALIJAGA YOGYAKARTA)

Bella Tasya Kumala Dewi, NIM.: 22206051010 (2025) KOMPARASI ALGORITMA RANDOM FOREST DAN NAIVE BAYES PADA INDOOR POSITIONING SYSTEM DENGAN METODE FINGERPRINTING (STUDI KASUS : GEDUNG KULIAH TERPADU UIN SUNAN KALIJAGA YOGYAKARTA). Masters thesis, UIN SUNAN KALIJAGA YOGYAKARTA.

Preview	Text (KOMPARASI ALGORITMA RANDOM FOREST DAN NAIVE BAYES PADA INDOOR POSITIONING SYSTEM DENGAN METODE FINGERPRINTING (STUDI KASUS : GEDUNG KULIAH TERPADU UIN SUNAN KALIJAGA YOGYAKARTA)) 22206051010_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf - Published Version Download (6MB) | Preview
	Text (KOMPARASI ALGORITMA RANDOM FOREST DAN NAIVE BAYES PADA INDOOR POSITIONING SYSTEM DENGAN METODE FINGERPRINTING (STUDI KASUS : GEDUNG KULIAH TERPADU UIN SUNAN KALIJAGA YOGYAKARTA)) 22206051010_BAB-II_sampai_SEBELUM-BAB-TERAKHIR.pdf - Published Version Restricted to Registered users only Download (7MB) | Request a copy

Abstract

The Global Positioning System (GPS) exhibits limited accuracy in indoor environments due to weak satellite signal penetration. Therefore, an alternative solution is needed in the form of an Indoor Positioning System (IPS). This study aims to design and analyze an IPS using a Wi-Fi signal-based fingerprinting method by comparing the performance of the Random Forest and Naïve Bayes algorithms and evaluating the impact of K-Means Clustering as a data preprocessing technique. Data were collected from the Integrated Lecture Building of UIN Sunan Kalijaga Yogyakarta using the “Cek Sinyal” application, which recorded parameters such as Received Signal Strength Indicator (RSSI), frequency, and environmental conditions. The dataset was subsequently processed through preprocessing, model training, and testing using evaluation metrics such as precision, recall, and f1-score. The results indicate that the Naïve Bayes algorithm achieved a very high f1-score (approaching 1.000) after the application of K-Means Clustering. However, the performance is likely affected by overfitting due to sensitivity to the clustered data distribution. Conversely, the Random Forest algorithm demonstrated more stable and consistent performance, achieving a maximum f1-score of 0.94, with greater resilience to signal fluctuations and noise. It is concluded that integrating fingerprinting methods with machine learning algorithms can produce an accurate and efficient IPS. The Random Forest algorithm is recommended for long-term implementation due to its superior generalization capability and reliability in dynamic real-world indoor environments.

Share this knowledge with your friends :

Actions (login required)

View Item