Pamukkale Univ Muh Bilim Derg. 2012; 18(3): 221-229 | DOI: 10.5505/pajes.2012.09719

Türkiye’de 1977-2006 Yýllarý Arasýnda Meydana Gelen Aylýk Trafik Kazalarýnýn Zamansal Analizi

Ahmet Atalay¹, Ahmet Tortum², Mahir Gökdað^2
1Atatürk Üniversitesi, Narman Meslek Yüksekokulu, Ýnþaat Bölümü, 25530, Narman, Erzurum
²Atatürk Üniversitesi, Mühendislik Fakültesi, Ýnþaat Mühendisliði Bölümü, 25240, Erzurum

Bu çalýþmada 1977-2006 yýllarý arasýnda meydana gelen aylýk trafik kaza verileri(þehir içi ve þehir dýþý toplamý) kullanýlarak zaman serisi analiz yöntemi ile modelleme yapýlmýþtýr. Yapýlan analizler sonucunda çalýþma döneminde kullanýlan verilere göre en uygun modelin ARIMA(4,1,4) olduðu belirlenmiþtir. Çalýþmada en uygun model kullanýlarak 2006: 01-2007: 12 dönemi için aylýk kaza tahmini yapýlmýþtýr. Tahmin edilen ve gerçek kaza deðerleri kullanýlarak regresyon eðrisi çizilmiþ ve korelasyon katsayýsý (r=0,9163) belirlenmiþtir. Tahmin deðerleri ile gerçek deðerler arasýnda güçlü pozitif iliþki olduðu belirlenmiþtir. Ayrýca ARIMA(4,1,4) modelinin baþarý ölçütü ortalama karesel hatalarýn karekökü (OKHK) deðeri hesaplanarak belirlenmiþtir. Çalýþma dönemi boyunca en fazla trafik kazasý Aralýk, Ekim ve Kasým aylarýnda, en az trafik kazasý Þubat, Mart ve Nisan aylarýnda meydana geldiði belirlenmiþtir.

Anahtar Kelimeler: Trafik kazalarý, Zamansal analiz, Zaman serisi analizi.

---

Temporal Analysis of Monthly Road Traffic Accidents Occurred in Turkey Between 1977-2006

Ahmet Atalay¹, Ahmet Tortum², Mahir Gökdað^2
1Atatürk University, Narman Vocational School, Civil Division, 25530, Narman, Erzurum
²Atatürk University, Faculty Of Engineering, Department Of Civil Engineering, 25240, Erzurum

In this study, modeling with time series analysis was done method by using date of monthly road traffic accidents (RTAs) occurred between1977-2006. According to results of analysis and date in period of this study, it was determined that ARIMA(4,1,4) was the best model. In the study monthly accident number are forecasted by appropriate model for 2006: 1-2007: 12 periods. Regression curve is drawn by using the estimated and actual number of RTAs and the correlation coefficient (r = 0.9163) were determined. A strong positive relationship was determined between estimated values and actual values. In addition, the success measure of ARIMA(4,1,4) model was identified by calculating value of root mean square errors (RMSE). It was determined that occur up to a traffic accident in December, October and November, at least a traffic accident in February, March and April

Keywords: Road traffic accidents, Temporal analysis, Time series analysis.

---

ATIF KOPYALA