A Multivariate Analysis of Crash and Naturalistic Driving Data in Relation to Highway Factors (2013) / Chapter Skim
Currently Skimming:
Pages 16-23
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 16... ...
16 Statistical Analysis: A Unified Approach to the Analysis of Rates for Crashes and Crash Surrogates Traditional Analysis of Crash Data In transportation-related safety studies, various data analytic methodologies have been used to investigate associations between crashes and various risk factors. Historically, depending on the application under investigation, Poisson, negative binomial, random effects, and hierarchical Bayesian data models, among others, have been used to analyze data collected from historical crash databases.
|
From page 17... ...
17 and independently fitting a separate model for a crash surrogate, the idea is to fit one model that accommodates both responses in a unifying model. The model is based on the method of seemingly unrelated regressions (SURs)
|
From page 18... ...
18 Then, the matrix S has dimension N × N Since this model satisfies the properties of a linear model with a defined covariance matrix, the parameters can be estimated by weighted least squares (WLS)
|
From page 19... ...
19 distribution to the desired posterior distributions. To ensure convergence, Markov chains are run with 60,000 iterations, and the first 30,000 are discarded for "burn-in." The Bayesian model has an important advantage over the classical model.
|
From page 20... ...
20 are indicated by the 2.5 and 97.5 percentiles. Log exposure is fit on the right-hand side of the model equations for both the crash and the surrogate regressions.
|
From page 21... ...
21 curve to no curve while holding other variables in the model fixed. The mean of this distribution is 0.38 with a 95% confidence interval of (0.15, 0.61)
|
From page 22... ...
22 Time to Edge Crossing Table 5.5 shows posterior estimates from the regression parameters in the TTEC model. In terms of regression estimates between the crash and surrogate measure, this model shows the best agreement among the three models.
|
From page 23... ...
23 intervention such as road widening, improving lane markings, changing signage or adding rumble strips can be evaluated by using the surrogate in a relatively short space of time. The effect on RR then represents a predicted safety benefit.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.