Analysis of Critical Factors Influencing Online Motorcycle Taxi Driver's Income Per Transaction Using Random Forest Regressor And Feature Importance
DOI:
https://doi.org/10.15575/istek.v14i2.2338Keywords:
Random Forest Regression, Driver Income, Feature Importance, Platform RatesAbstract
This study aims to identify and measure the main factors that most significantly affect the Income of Online Motorcycle Taxi Drivers Per Transaction in the gig economy sector. The Machine Learning Random Forest Regressor algorithm was used on driver transaction data. This methodology was chosen for its ability to handle the data's non-linearity and to objectively measure Feature Importance. Traditional linear regression models have limitations in these areas. The main results show the Random Forest model is highly accurate (R2 = 0.9634). It confirms the absolute dominance of distance, which accounts for 94.98% of the total predictive importance of revenue. The Total Transaction Value factor (3.82%) is a secondary predictor. Demographic variables (Age and Gender) and temporal variables (Days and Hours) together had a minimal (less than 1%) influence on fare per trip. This research concludes that the rate per driver transaction is determined almost exclusively by the platform's distance-based pricing policy. It is neutral to the characteristics of the driver. These findings recommend that platforms focus on increasing order volume and optimizing operational costs, rather than modifying base rates.
References
[1] S. J. Hong, J. M. Bauer, K. Lee, dan N. F. Granados, “Drivers of Supplier Participation in Ride-Hailing Platforms,” Journal of Management Information Systems, vol. 37, no. 3, hlm. 602–630, Jul 2020, doi: 10.1080/07421222.2020.1790177.
[2] S. Darmastuti, A. Rahmawati, R. L. Putri, dan A. Sundusiyah, “The Rise of the Gig Economy: Job Creation and the Paradox of Working Relationship on the Digital Transportation Platform.”
[3] M. Li, X. Hu, K. Jin, dan J. Han, “Exploring factors influencing entry into the gig economy: A study of Chinese workers,” Acta Psychol (Amst), vol. 259, Sep 2025, doi: 10.1016/j.actpsy.2025.105301.
[4] J. Woodcock dan M. Graham, “A Critical Introduction.”
[5] E. Mariano, V. Amkeun2, N. Bau3, dan Y. P. Lian, “Analisis Pendapatan Pengemudi Ojek Online Maxim-Bike,” vol. 2, 2023, [Daring]. Tersedia pada: http://jurnal.jomparnd.com/index.php/jk
[6] D. S. Pratomo, P. M. A. Saputra, D. A. N. Asrofi, C. Natalia, dan S. L. Zenritami, “Gig Workers In The Digital Era In Indonesia: Development, Vulnerability And Welfare,” 2024, hlm. 47–60. doi: 10.2991/978-94-6463-525-6_6.
[7] L. Breiman, “Random Forests,” 2001.
[8] Y. Yang dan H. Wang, “Random Forest-Based Machine Failure Prediction: A Performance Comparison,” Applied Sciences (Switzerland), vol. 15, no. 16, Agu 2025, doi: 10.3390/app15168841.
[9] T. Tulabandhula dan C. Rudin, “Machine Learning with Operational Costs,” 2013.
[10] K. R. Putra, “Comparison of Prediction Models: Decision Tree, Random Forest, and Support Vector Regression,” vol. 6, no. 1, hlm. 39–49, 2025, doi: 10.33365/jatika.v6i1.18.
[11] E. Bhaduri dan S. Pal, “Bhaduri, Pal, and Goswami Analysing Factors Affecting the Adoption of Ride-Hailing Services (RHS) in India? A Step-Wise LCCA-MCDM Modeling Approach.”
[12] X. Chen, S. Bai, Y. Wei, dan H. Jiang, “How income satisfaction impacts driver engagement dynamics in ride-hailing services,” Transp Res Part C Emerg Technol, vol. 157, hlm. 104418, Des 2023, doi: 10.1016/j.trc.2023.104418.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Agung Wahana, Cecep Nurul Alam
This work is licensed under a Creative Commons Attribution 4.0 International License.