Bambang Haryadi


Abstract: The objectives of the presents study was  to explore mathematical models that could be used  to  estimate  the  number  of  accidents  on  interurban  toll  roads.  Preditive  models  were developed  by  relating  traffic  exposure  (average  daily  traffic  and  road  section  length)  with  the number  of  accidents  per  unit  of  time.Accident,  roadway  section  and  traffic  volume  data  were obtained  from  Jagorawi,  JakartaCikampek,  padaleunyi,  and  palikanci  toll  road.  Accident  rate models were developed from those data using  generalized linear modelling (GLM) techniques. The conclusions  from  the  study were:  (1)  linier  regression model was  not  appropriate  to  be  used  to predict  accidents  number,  because  accident  occurence  did  not  follow  normal  distribution,  (2) Poisson regression possessed accident occurence characteristics: descrete, rare and random, and (3)  Negative  Binomial  distribution  was  more  appropriate  to  represent  accident  occurence phenomenon with overdispersion.

Keywords: accident rate, generalized linear modelling, Poisson regression, nagative binomial regression.


Abstrak: Penelitian  ini  bertujuan  untuk mengeksplorasi model matematis  yang  dapat  digunakan untuk  meramalkan  jumlah  kecelakaan,  pada  suatu  ruas  jalan  tol  antar  kota.  Model  prediksi dikembangkan  dengan  menghubungkan  paparan  lalulintas,  yang  dinyatakan  dengan  volume lalulintas dan panjang ruas jalan, dengan keselamatan, yang dinyatakan dalam jumlah kecelakaan per satuan waktu. Data  jalan, kecelakaan, dan  lalulintas selama dua  tahun, diambil dari  jalan  tol antar  kota  Jagorawi,  Jakarta0Cikampek,  Padaleunyi  dan  Palikanci.  Model  tingkat  kecelakaan dikembangkan dari data tersebut dengan teknik generalized linear modelling (GLM), dan dikalibrasi dengan  menggunakan  teknik0teknik  statistik.  Beberapa  kesimpulan  yang  dapat  ditarik  dari penelitian  ini  adalah:  (1)  Model  regresi  linier  dengan  distribusi  normal  tidak  memadai  untuk memprediksi  kecelakaan;  tingkat  kecelakaan  tidak  terbukti  mengikuti  distribusi  normal;  (2) Penggunaan  regresi Poisson  lebih mencerminkan karakteristik kejadian kecelakaan: diskrit, acak, langka;    dan  (3)  Distribusi  Binomial  Negatif  paling  baik merepresentasikan  kejadian  kecelakaan dengan adanya gejala overdispersi pada data.

Kata kunci: tingkat kecelakaan, generalized linear modelling, regresi poisson, regresi binomial negatif.


accident rate; generalized linear modelling; Poisson regression; nagative binomial regression; tingkat kecelakaan; generalized linear modelling; regresi poisson; regresi binomial negatif

Full Text:

PDF PostScript


Allison, Paul D. 1999. Logistic Regression Using the SAS System: Theory and Application. Cary, NC.: AS Institute Inc.

Bonneson, J. A., and P. T. McCoy. 1997. “Effect of Median Treatment on Urban Arterial Safety: An Accident Prediction Model,” Highway Research Record 1581, Highway Research Board, Washington, D.C.

Brown, H. C. and A. P. Tarko. 1999. “Effects of Access Control on Safety on Urban Arterial Streets,” Highway Research Record 1665, Highway Research Board, Washington, D.C.

Devore, Jay L., 1991. Probability and Statistics for Engineering and the Sciences. Brooks/Cole Publishing Company, Pasific Grove, California.

Garber, N.J. and A.A. Ehrhart, 2000. The effects of speed, flow, and geometric characteristics on crash rates for different types of virginia highways. Final Report, Virginia Transportation Research Council, VTRC 000R15.

Hadi, M., J. Aruldhas, L.F. Chow, and J.J.A. Wattleworth, 1995. Estimating safety effects of cross0section design of various hyghway types using negative binomial regression. Transport. Res. Record. 1500:1690176.

Knuiman, M. W., Council, F.M. and D. W. Reinfurt. 1993. “Association of Median Width and Highway Accident Rates,” Highway Research Record 1401, Highway Research Board, Washington, D.C.

Lord, D. 2002. Application of Accident Prediction Models for Computation of Accident Risk on Transportation Network, Highway Research Records 1784, Highway Research Board, Washington, D.C.

Lord, D., and B. N. Persaud. 2000. “Accident Prediction Models With and Without Trend: Application of the Generalized Estimating Equations Procedure,” Highway Research Record 1717, Highway Research Board, Washington, D.C.

Maher, M.J. and I. Summersgill, 1996. A comparative methodology for the fitting of predictive accident models. Accid. Anal. And Prev., 28(3): 2810296.

Miaou, S., Hu, P. S., Wright, T., Rathi, A. K. and S. C. Davis. 1992. “Relationship Between Truck Accidents and Highway Geometric Design: A Poisson Regression Approach,” Highway Research Record 1376, Highway Research Board, Washington, D.C.

Neter, J., W. Wasserman, and M.H. Kutner, 1999. Applied Regression Models, 2nd ed.Richard D. Irwin, Inc., Boston.

Petruccelli, Joseph D, B. Nandram and M. Chen. 1999. Applied Statistics for Engineers and Scientists. New Jersey; Prentice Hall.

Qin, X; Ivan, J.N.; Ravishanker, N., & Liu, J. 2005. Hierarchical Bayesian Estimation of Safety Performance Functions for Two0Lane Highways Using Markov chain Monte Carlo Modeling, Journal of Transportation Engineering, Vol. 131, No. 5.

Rice, John A. 1988. Mathematical Statistics and Data Analysis. California; Wadsworth & Brooks/Cole Advanced Books & Software.

Saccomano, F.F. & C. Buyco. 1988. Generalized Loglinear Models of Truck Accident Rates, Highway Research Record 1172, Highway Research Board, Washington, D.C.

DOI: https://doi.org/10.15294/jtsp.v13i1.1338


  • There are currently no refbacks.