A Hybrid System With K-Nearest Neighbor (KNN) and MLP Method: Traffic Flow Estimation

Year 2024, Volume: 26 Issue: 4, 1801 - 1816, 17.12.2024

Yavuz Selim Balcıoğlu , Bülent Sezen

https://doi.org/10.32709/akusosbil.1255897

Abstract

In this research, we offer a hybrid traffic flow prediction approach that combines k-nearest neighbor (KNN) and multilayer perceptron (MLP). This model is referred to as the KNN-MLP model. The goal of this method is to increase the accuracy of the predictions. KNN is used to pick surrounding stations that are mostly connected to the test station and to capture the spatial characteristics of traffic flow. The Multi-Layer Perceptron (MLP) algorithm is used to mine the temporal variability of traffic flow, and a four-layer MLP network is used to forecast traffic flow correspondingly at chosen stations. The result-level fusion combined with the rank-exponent weighting approach is used to get the final prediction results. The accuracy of the forecast is determined using data on the current flow of traffic that is collected in real time by the Department of Transportation Data Center of the Istanbul Metropolitan Municipality. According to the findings of the experiments, the proposed model has the potential to achieve a higher level of performance when compared to well-known prediction models such as support vector regression (SVR), LSTM, and MLP models. Furthermore, the proposed model has the potential to achieve an improvement in accuracy of approximately 2% on average.

Keywords

prediction, traffic, knn, mlp, machine learning

K- En Yakın Komşu (KNN) ve MLP Yöntemi ile Hibrit Bir Sistem: Trafik Akış Tahmini

Abstract

Bu araştırmada, k-en yakın komşu (KNN) ve çok katmanlı algılayıcıyı (MLP) birleştiren hibrit bir trafik akışı tahmin yaklaşımı sunuyoruz. Bu model KNN-MLP modeli olarak adlandırmaktadır. Bu yöntemin amacı tahminlerin doğruluğunu arttırmaktır. KNN, çoğunlukla test istasyonuna bağlı olan çevredeki istasyonları seçmek ve trafik akışının mekansal özelliklerini yakalamak için kullanılır. Trafik akışının zamansal değişkenliğini araştırmak için Çok Katmanlı Algılayıcı (MLP) algoritması kullanılmış ve seçilen istasyonlarda buna uygun olarak trafik akışını tahmin etmek için dört katmanlı bir MLP ağı kullanılmıştır. Nihai tahmin sonuçlarını elde etmek için sıra-üs ağırlıklandırma yaklaşımıyla birleştirilmiş sonuç düzeyinde füzyon kullanılmıştır. Tahminin doğruluğu, İstanbul Büyükşehir Belediyesi Ulaşım Daire Başkanlığı Veri Merkezi tarafından gerçek zamanlı olarak toplanan mevcut trafik akışı verileri kullanılarak belirlenmiştir. Deneylerden elde edilen bulgulara göre, önerilen model destek vektör regresyon (SVR), LSTM ve MLP modelleri gibi bilinen tahmin modellerine göre daha yüksek performans düzeyine ulaşma potansiyeline sahiptir. Ayrıca, önerilen modelin doğruluğunda ortalama olarak yaklaşık %2'lik bir iyileşme sağlanmıştır.

Keywords

trafik, tahmin, knn, mlp, makine öğrenmesi

References

• Duan, Y. et al. (2016) ‘An efficient realization of deep learning for traffic data imputation’, Transportation Research Part C: Emerging Technologies. doi: 10.1016/j.trc.2016.09.015.
• Farid, A., Abdel-Aty, M. and Lee, J. (2019) ‘Comparative analysis of multiple techniques for developing and transferring safety performance functions’, Accident Analysis and Prevention. doi: 10.1016/j.aap.2018.09.024.
• Finogeev, A. et al. (2019) ‘Intelligent monitoring system for smart road environment’, Journal of Industrial Information Integration. doi: 10.1016/j.jii.2019.05.003.
• He, Y., Tablada, A. and Wong, N. H. (2019) ‘A parametric study of angular road patterns on pedestrian ventilation in high-density urban areas’, Building and Environment. doi: 10.1016/j.buildenv.2019.01.047.
• Hoang, N. D. and Nguyen, Q. L. (2018) ‘Automatic Recognition of Asphalt Pavement Cracks Based on Image Processing and Machine Learning Approaches: A Comparative Study on Classifier Performance’, Mathematical Problems in Engineering. doi: 10.1155/2018/6290498.
• Huang, W. et al. (2014) ‘Deep architecture for traffic flow prediction: Deep belief networks with multitask learning’, IEEE Transactions on Intelligent Transportation Systems. doi: 10.1109/TITS.2014.2311123.
• Jiang, R. et al. (2019) ‘Deepurbanevent: A system for predicting citywide crowd dynamics at big events’, in Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. doi: 10.1145/3292500.3330654.
• Jiber, M. et al. (2020) ‘Road traffic prediction model using extreme learning machine: The case study of tangier, morocco’, Information (Switzerland). doi: 10.3390/info11120542.
• Koesdwiady, A., Soua, R. and Karray, F. (2016) ‘Improving Traffic Flow Prediction with Weather Information in Connected Cars: A Deep Learning Approach’, IEEE Transactions on Vehicular Technology. doi: 10.1109/TVT.2016.2585575.
• Li, L. et al. (2019) ‘Day-ahead traffic flow forecasting based on a deep belief network optimized by the multi-objective particle swarm algorithm’, Knowledge-Based Systems. doi: 10.1016/j.knosys.2019.01.015.
• Lv, Y. et al. (2015) ‘Traffic Flow Prediction with Big Data: A Deep Learning Approach’, IEEE Transactions on Intelligent Transportation Systems. doi: 10.1109/TITS.2014.2345663.
• Polson, N. and Sokolov, V. (2016) ‘Deep Learning Predictors for Traffic Flows’, arXiv stat.AP.
• Pompigna, A. and Mauro, R. (2022) ‘Smart roads: A state of the art of highways innovations in the Smart Age’, Engineering Science and Technology, an International Journal. doi: 10.1016/j.jestch.2021.04.005.
• Sahitya, K. S. and Prasad, C. S. R. K. (2021) ‘GIS-based urban road network accessibility modeling using MLR, ANN and ANFIS methods’, Transport and Telecommunication. doi: 10.2478/ttj-2021-0002.
• Satti, S. K. et al. (2021) ‘A machine learning approach for detecting and tracking road boundary lanes’, ICT Express. doi: 10.1016/j.icte.2020.07.007.
• Song, M. and Civco, D. (2004) ‘Road extraction using SVM and image segmentation’, Photogrammetric Engineering and Remote Sensing. doi: 10.14358/PERS.70.12.1365.
• Soua, R., Koesdwiady, A. and Karray, F. (2016) ‘Big-data-generated traffic flow prediction using deep learning and dempster-shafer theory’, in Proceedings of the International Joint Conference on Neural Networks. doi: 10.1109/IJCNN.2016.7727607.
• Oliveira, D. D., Rampinelli, M., Tozatto, G. Z., Andreão, R. V., & Müller, S. M. (2021). Forecasting vehicular traffic flow using MLP and LSTM. Neural Computing and applications, 33, 17245-17256.
• Vlahogianni, E. I., Karlaftis, M. G., & Golias, J. C. (2005). Optimized and meta-optimized neural networks for short-term traffic flow prediction: A genetic approach. Transportation Research Part C: Emerging Technologies, 13(3), 211-234.
• Wang, J. and Boukerche, A. (2021) ‘Non-parametric models with optimized training strategy for vehicles traffic flow prediction’, Computer Networks. doi: 10.1016/j.comnet.2020.107791.
• Xu, D. et al. (2020) ‘Real-time road traffic state prediction based on kernel-KNN’, Transportmetrica A: Transport Science. doi: 10.1080/23249935.2018.1491073.
• Yang, L. et al. (2020) ‘Griffin: An Ensemble of AutoEncoders for Anomaly Traffic Detection in SDN’, in 2020 IEEE Global Communications Conference, GLOBECOM 2020 - Proceedings. doi: 10.1109/GLOBECOM42002.2020.9322187.
• Zargari, S. A., Siabil, S. Z., Alavi, A. H., & Gandomi, A. H. (2012). A computational intelligence‐based approach for short‐term traffic flow prediction. Expert Systems, 29(2), 124-142.
• Zhang, H. et al. (2020) ‘A real-Time and ubiquitous network attack detection based on deep belief network and support vector machine’, IEEE/CAA Journal of Automatica Sinica. doi: 10.1109/JAS.2020.1003099.
• Zhang, L. et al. (2013) ‘An Improved K-nearest Neighbor Model for Short-term Traffic Flow Prediction’, Procedia - Social and Behavioral Sciences. doi: 10.1016/j.sbspro.2013.08.076.
• Zhao, Z. et al. (2017) ‘LSTM network: A deep learning approach for Short-term traffic forecast’, IET Intelligent Transport Systems. doi: 10.1049/iet-its.2016.0208.
• Zhou, T. et al. (2017) ‘δ-agree AdaBoost stacked autoencoder for short-term traffic flow forecasting’, Neurocomputing. doi: 10.1016/j.neucom.2017.03.049.