Analisis Prediksi Harga Mobil Bekas di Pasar Indonesia Menggunakan Metode Support Vector Regression (SVR)

Ismail, Farellino Azhfar Ismail (2026) Analisis Prediksi Harga Mobil Bekas di Pasar Indonesia Menggunakan Metode Support Vector Regression (SVR). Diploma thesis, Institut Teknologi Sepuluh Nopember.

Text 2043221121-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Transportasi berperan penting dalam mendukung aktivitas ekonomi dan sosial masyarakat. Mobil menjadi salah satu moda transportasi yang banyak digunakan karena memberikan kenyamanan dan fleksibilitas dalam mobilitas sehari-hari. Di Indonesia, permintaan terhadap mobil bekas terus meningkat karena dinilai lebih ekonomis dibandingkan mobil baru. Namun, penentuan harga mobil bekas sering kali tidak didasarkan pada kondisi kendaraan secara objektif, melainkan hanya berdasarkan intuisi atau tren pasar, sehingga menimbulkan kesenjangan informasi antara penjual dan pembeli. Penelitian ini bertujuan membangun model Support Vector Regression (SVR) untuk memprediksi harga mobil bekas Toyota berdasarkan karakteristik kendaraan, sehingga dapat membantu memberikan estimasi harga yang lebih objektif dan akurat. Penelitian ini membandingkan kinerja beberapa fungsi kernel pada algoritma SVR, yaitu Linear, Radial Basis Function (RBF), Polynomial, dan Sigmoid. Data yang digunakan merupakan hasil web scraping dari website OLX Indonesia sebanyak 3.305 data mentah, kemudian melalui proses preprocessing yang meliputi pembersihan data, penanganan missing value, penghapusan outlier, normalisasi, dan transformasi variabel sehingga diperoleh 1.711 data bersih yang digunakan dalam proses pemodelan. Optimasi hyperparameter dilakukan menggunakan Grid Search dengan 5-fold Cross Validation, sedangkan evaluasi model menggunakan metrik Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), dan Mean Absolute Percentage Error (MAPE). Hasil penelitian menunjukkan bahwa model SVR dengan kernel RBF memberikan performa terbaik dibandingkan kernel lainnya. Model tersebut memperoleh nilai MAPE sebesar 8,70%, MAE sebesar Rp23.750.625, dan RMSE sebesar Rp36.251.250 dengan parameter optimal C = 10, epsilon = 0,01, dan gamma = auto. Hasil tersebut menunjukkan bahwa metode SVR dengan kernel RBF memiliki tingkat akurasi yang sangat baik dalam memprediksi harga mobil bekas Toyota, sehingga dapat dimanfaatkan sebagai solusi yang lebih objektif dalam membantu penjual maupun pembeli menentukan nilai wajar kendaraan secara transparan.
=================================================================================================================================
Transportation plays an important role in supporting economic and social activities. Used cars have become increasingly popular in Indonesia because they offer a more affordable alternative to new vehicles. However, used car prices are often determined subjectively based on intuition or market trends rather than the actual condition of the vehicle, resulting in information asymmetry between sellers and buyers. This study aims to develop a Support Vector Regression (SVR) model to predict the prices of used Toyota cars based on their vehicle characteristics, thereby providing more objective and accurate price estimations. This study compares the performance of several kernel functions in the SVR algorithm, namely Linear, Radial Basis Function (RBF), Polynomial, and Sigmoid kernels. The dataset consisted of 3,305 raw records collected through web scraping from the OLX Indonesia website. After preprocessing, including data cleaning, missing value handling, outlier removal, normalization, and variable transformation, 1,711 valid records were obtained for model development. Hyperparameter optimization was performed using Grid Search with 5-fold Cross Validation, while model performance was evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE). The results indicate that the SVR model with the RBF kernel outperformed the other kernels. The best-performing model achieved a MAPE of 8.70%, an MAE of Rp23,750,625, and an RMSE of Rp36,251,250, with the optimal hyperparameters of C = 10, epsilon = 0.01, and gamma = auto. These findings demonstrate that the SVR method with the RBF kernel provides excellent predictive accuracy for estimating used Toyota car prices and can serve as an objective decision-support tool for both sellers and buyers in determining fair vehicle prices.

Item Type:	Thesis (Diploma)
Uncontrolled Keywords:	Harga, Mobil Bekas, Machine Learning, Pasar, Support Vector Regression, Support Vector Machine
Subjects:	Q Science Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Vocational > 49501-Business Statistics
Depositing User:	Farellino Azhfar Ismail
Date Deposited:	30 Jul 2026 06:56
Last Modified:	30 Jul 2026 06:56
URI:	http://repository.its.ac.id/id/eprint/139630

Actions (login required)

View Item