PENGEMBANGAN SISTEM PREDIKSI RATING PROGRAM TV BERBASIS DASHBOARD MENGGUNAKAN MACHINE LEARNING UNTUK MENDUKUNG PENGAMBILAN KEPUTUSAN (STUDI KASUS: GARUDA TV)

Michael Steven, - and Liptia Venica, - and Muhammad Rizalul Wahid, - (2025) PENGEMBANGAN SISTEM PREDIKSI RATING PROGRAM TV BERBASIS DASHBOARD MENGGUNAKAN MACHINE LEARNING UNTUK MENDUKUNG PENGAMBILAN KEPUTUSAN (STUDI KASUS: GARUDA TV). S1 thesis, Universitas Pendidikan Indonesia.

Abstract

Garuda TV masih mengandalkan metode konvensional dalam menganalisis dan memprediksi rating program televisi, yang menyebabkan keterlambatan laporan serta kurang optimalnya pengambilan keputusan strategis. Oleh karena itu, diperlukan solusi berbasis machine learning yang mampu menghasilkan prediksi secara cepat dan akurat. Penelitian ini menggunakan pendekatan Design Science Research Methodology (DSRM) untuk mengembangkan model prediksi rating dan dashboard analitik pendukung keputusan. Tiga algoritma diuji, yaitu Random Forest, XGBoost, dan Support Vector Machine (SVM), menggunakan dataset historis rating tujuh program utama Garuda TV periode September 2024–Februari 2025, dengan tambahan data Maret 2025 untuk evaluasi lanjutan. Tahapan penelitian meliputi data cleaning, seleksi fitur, hyperparameter tuning, serta evaluasi menggunakan metrik RMSE, MAE, MAPE, dan R². Hasil pengujian menunjukkan bahwa XGBoost memberikan performa terbaik dan paling stabil dibandingkan algoritma lainnya. Sebagai contoh, pada program Laporan 8 Siang, model XGBoost setelah penambahan data Maret menghasilkan nilai RMSE sebesar 0,0052, MAE sebesar 0,0039, MAPE sebesar 0,13%, serta R² sebesar 0,8961 yang menunjukkan tingkat akurasi dan kemampuan generalisasi yang tinggi. Selain itu, dashboard yang dikembangkan mampu menampilkan prediksi rating secara real-time, visualisasi performa program, serta rekomendasi strategis, dan telah dinyatakan dapat diterima serta memiliki tingkat kebergunaan yang baik berdasarkan hasil User Acceptance Testing (UAT) dan System Usability Scale (SUS). Secara keseluruhan, penerapan machine learning, khususnya XGBoost, terbukti efektif dalam meningkatkan akurasi prediksi rating dan mendukung pengambilan keputusan strategis di Garuda TV. ----- Garuda TV still relies on conventional methods to analyze and predict television program ratings, which leads to delays in reporting and less optimal strategic decision-making. Therefore, a machine learning–based solution is required to generate faster and more accurate rating predictions. This study employs the Design Science Research Methodology (DSRM) to develop a machine learning–based rating prediction model and an analytics dashboard as a decision support system. Three algorithms were evaluated, namely Random Forest, XGBoost, and Support Vector Machine (SVM), using historical rating data from seven major Garuda TV programs covering the period from September 2024 to February 2025, with additional data from March 2025 included for further evaluation. The research process involved data cleaning, feature selection, hyperparameter tuning, and model evaluation using RMSE, MAE, MAPE, and R² as performance metrics. The experimental results indicate that XGBoost achieves the best and most stable performance compared to the other algorithms. For instance, for the Laporan 8 Siang program, the XGBoost model after incorporating the March dataset produced an RMSE of 0.0052, MAE of 0.0039, MAPE of 0.13%, and an R² value of 0.8961, demonstrating high accuracy and strong generalization capability. Furthermore, the developed dashboard is capable of presenting real-time rating predictions, program performance visualizations, and strategic recommendations. The system was evaluated using User Acceptance Testing (UAT) and the System Usability Scale (SUS), and the results indicate that it is acceptable and exhibits a good level of usability. Overall, the application of machine learning, particularly the XGBoost algorithm, has proven effective in improving rating prediction accuracy and supporting strategic decision-making at Garuda TV.

	Text S_MKB_2100210_Title.pdf Download (449kB)
	Text S_MKB_2100210_Chapter1.pdf Download (292kB)
	Text S_MKB_2100210_Chapter2.pdf Restricted to Staf Perpustakaan Download (753kB)
	Text S_MKB_2100210_Chapter3.pdf Download (1MB)
	Text S_MKB_2100210_Chapter4.pdf Restricted to Staf Perpustakaan Download (1MB)
	Text S_MKB_2100210_Chapter5.pdf Download (261kB)
	Text S_MKB_2100210_Appendix.pdf Restricted to Staf Perpustakaan Download (509kB)

Item Type:	Thesis (S1)
Additional Information:	https://scholar.google.com/citations?user=SV0-9c4AAAAJ&hl=id&authuser=2 ID SINTA DOSEN PEMBIMBING : Liptia Venica : 6779029 Muhammad Rizalul Wahid : 6780434
Uncontrolled Keywords:	Prediksi Rating TV, Machine Learning, Random Forest, XGBoost, Support Vector Machine TV Rating Prediction, Machine Learning, Random Forest, XGBoost, Support Vector Machine
Subjects:	T Technology > T Technology (General)
Divisions:	UPI Kampus Purwakarta > S1 Mekatronika dan Kecerdasan Buatan
Depositing User:	Michael Steven
Date Deposited:	17 Dec 2025 08:59
Last Modified:	17 Dec 2025 08:59
URI:	http://repository.upi.edu/id/eprint/145888

Actions (login required)

View Item