TRAFFIC VOLUME FORECAST USING REGRESSION ANALYSIS AND ARTIFICIAL NEURAL NETWORK BASED ON PRINCIPAL COMPONENTS

Ramadan Duraku; Vaska Atanasova

Scientific paper ID 1644 : 2018/3

TRAFFIC VOLUME FORECAST USING REGRESSION ANALYSIS AND ARTIFICIAL NEURAL NETWORK BASED ON PRINCIPAL COMPONENTS

Ramadan Duraku, Vaska Atanasova

Aim of this Paper is to explore which factors have the greatest impact on generating traffic volume and to set up a suitable model for forecasting it for the main road network of Anamorava region. In this regard, several demographic and social economic variables were taken into account for the period 2004-2016. Variability between variables is done through correlative analysis (multi-co linearity problem) between them. After this, models were developed using original data via multiple regression analysis (MLR) and artificial neural network (ANN) methods. However, with the aim of involving as many variables as possible and eliminating the multi co-linearity respectively improving forecasting model capability, a new methodology was used using the principal component analysis (PCA) as an input. These models then are compared based on prediction errors also verified on the basis of some performance indicators. Results show that the use of principal components as inputs improved both models forecasting by reducing their complexity and eliminating data co-linearity.

• Reload article • Open/Download as PDF

обем на превозите модел прогнозиране основни компоненти множествена регресия изкуствена невронна мрежа.traffic volume model forecasting principal component multiple regression analysis artificial neural network.Ramadan Duraku Vaska Atanasova

BIBLIOGRAPHY

[1] Ministry of Infrastructure-Directorate of Roads. „Data for traffic volumes for the period 2004-2016“, Prishtina,2016

[2] Mihailo M., Planiranje Saobracaj i Prostora. Gredevinski Fakultet, Beograd, 2004

[3] Gjevori SH., Traffic Engineering-First Part (Book in Albanian). Polytechnic University of Tirana, Faculty of Mechanical Engineering, Tirana, 2011

[4] Neveu, A.J., Quick response procedure to forecast rural traffic. Transportation research record 944, June 1982

[5] Gjevori SH., Transport Systems. (Book in Albanian). Polytechnic University of Tirana-Faculty of Mechanical Engineering. Tirana, 2010

[6] Hutchinson, B.G. Principle of urban transport system planning. Script Book Company, Washington D.C.

[7] Agency Statistics of Kosovo. Kosovo Census Atlas, Geographic and Administrative Division of Kosovo. Prishtina, 2013.

[8] Agency Statistics of Kosovo,”General Statistics of Kosovo”.Prishtine, 2016.

[9] Slipunas T., Annual average daily traffic forecasting using different techniques. Transport and Road Research Institute. VolXXI,No.1,38-43, Lithuania, 2006.

[10] Mirko, Č. Modeliranje u Železnickom Saobračaju. Saobracajni Fakultet-Universitet u Beogradu. Beograd, 2003

[11] Washington S.P, Karlaftis M.G, Mannering F.L.,”Statistical and Econometric methods for transportation data”. Second Edtion, Taylor and Francis Group, LLC, 2011

[12] Sousa et al., ”Multiple linear regression and ANN based on principal components to predict ozone concentrations”. Elsevier, 2006

[13] Živojinović, D. Z. Razvoj i Primena Hemometrijskih Metoda za Klasifikaciju i Procenu Kvaliteta Vode. Doktorska Disertacija. Univerzitet u Beogradu-Tehnološko-Metalurški Fakultet. Beograd.2013.

[14] P.B Mistry, “Principal regression for crop yield estimation”. Springer, 2016.

[15] Ali, G.A., Awdalla, T. Characteristics and predictions of traffic accident causalities in Sudan using statistical modelling and ANN. IJTST, Vol 1. N0.4,2012-pg.305-317.

[16] Fricker J.D, Saha K.Sunil,. Traffic volume forecasting methods for rural state highways-Final report. FHWA/IN/JHRP-86.20, JHRP, Purdue University, 1987

[17] Long range Forecasting From Crystal Ball to Computer. www.forecastingprinciples.com/files/LRF-ch13.pdf