ESTIMATION OF PASSENGER-KILOMETER AND TONNEKILOMETER VALUES FOR HIGHWAY TRANSPORTATION IN TURKEY USING THE FLOWER POLLINATION ALGORITHM

KORKMAZ, Ersin; AKGUNGOR, Ali Payidar · 2018 · Crossref

DOI: 10.20858/sjsutst.2018.98.5

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the need for accurate forecasting of highway transportation demand in Turkey, where road transport accounts for approximately 90% of passenger and freight movements. Accurate estimation of passenger-kilometer and tonne-kilometer values is critical for developing effective transportation plans and policies to mitigate issues such as traffic accidents. The authors aim to demonstrate the applicability of the Flower Pollination Algorithm (FPA), an artificial intelligence optimization technique, in developing simple and practical demand forecasting models. The methodology involves developing three distinct model forms—linear, power, and semi-quadratic—to estimate passenger-kilometer and tonne-kilometer values. These models utilize three independent input parameters obtained from the Turkish Statistical Institute for the period between 1990 and 2016: population (P), gross domestic product per capita (GDPperC), and the number of vehicles (V). For passenger models, vehicle counts included cars, buses, and minibuses, while freight models utilized trucks and vans. The dataset was split into 22 years of training data and 5 years of test data. The FPA was employed to optimize the coefficients of these models, leveraging its ability to balance global and local search processes inspired by plant pollination behaviors. The results indicate that the semi-quadratic model provided the most accurate estimates, achieving the lowest Mean Absolute Percentage Error (MAPE) and highest coefficient of determination (R²) for both passenger and freight forecasts. Specifically, the semi-quadratic passenger-kilometer model achieved a MAPE of 3.09% on training data and 3.7% on test data, with R² values of 97.74%. The linear model served as a viable alternative due to its simplicity, whereas the power model performed poorly, yielding errors around 9%. Using the optimized semi-quadratic model, the authors projected demand for 2030 under two scenarios. Scenario 1 assumed higher growth rates in population, GDP, and vehicles, while Scenario 2 relied on official statistical institute projections. Both scenarios predicted a significant increase in transport demand, with Scenario 1 leading to an approximate 50% increase in demand due to higher input parameter values. The study concludes that the FPA is an effective tool for optimizing transportation demand models and that demand for both passenger and freight transport in Turkey will rise in parallel with population growth and increased prosperity. The findings suggest that the semi-quadratic model is the most robust for estimation, though the linear model remains a practical option. The authors highlight the potential for FPA to be applied in various other forecasting areas and propose future comparisons with other artificial intelligence techniques.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• exposure measurement