Distribution model choice, parameters’ estimation and measures of goodness of fit were carried out, and the best distribution fitting for incident duration data was LogLogistic distribution. Accelerated failure time (AFT) modeling was employed based on survival analysis and LogLogistic distribution. Then stepwise by Akaike information criterion (AIC) was used to find the best group of factors for the prediction of freeway incident duration. The test result indicates that AFT model can represent actual incident duration and accept different distribution and missing data.
Reference
Related
Cited by
Get Citation
CONG Haozhe, WANG Junhua, FANG Shouen, TONG Shixin. Fitting Distribution of Freeway Incident Duration and Accelerated Failure Time Modeling[J].同济大学学报(自然科学版),2011,39(10):1468~1472