Incorporating Temporal Effect into Crash Safety Performance Functions Wen Cheng, Ph.D., P.E., PTOE Civil Engineering Department Cal Poly Pomona.

Slides:



Advertisements
Similar presentations
Trisha Muñoz, E.I.T Civil Engineering Department Cal Poly Pomona.
Advertisements

Lesson 10: Linear Regression and Correlation
Kin 304 Regression Linear Regression Least Sum of Squares
LOGO ITE Presentation June 27, 2012 Presenter:Jung-Han Wang.
Probability & Statistical Inference Lecture 9
The General Linear Model Or, What the Hell’s Going on During Estimation?
Random effects estimation RANDOM EFFECTS REGRESSIONS When the observed variables of interest are constant for each individual, a fixed effects regression.
Spring  Crash modification factors (CMFs) are becoming increasing popular: ◦ Simple multiplication factor ◦ Used for estimating safety improvement.
Department of Civil Engineering University of Washington Quantitative Safety Analysis for Intersections on Washington State Two-lane Rural Highways Master’s.
Investigation of Varied Time Intervals in Crash Hotspot Identification Authors: Wen Cheng, Ph.D., P.E., Fernando Gonzalez, EIT, & Xudong Jia; California.
LINEAR REGRESSION: Evaluating Regression Models Overview Assumptions for Linear Regression Evaluating a Regression Model.
Spring Sampling Frame Sampling frame: the sampling frame is the list of the population (this is a general term) from which the sample is drawn.
LINEAR REGRESSION: Evaluating Regression Models. Overview Assumptions for Linear Regression Evaluating a Regression Model.
Yard. Doç. Dr. Tarkan Erdik Regression analysis - Week 12 1.
Multivariate Data Analysis Chapter 4 – Multiple Regression.
Summarizing Empirical Estimation EconS 451: Lecture #9 Transforming Variables to Improve Model Using Dummy / Indicator Variables Issues related to Model.
Incorporating Safety into the Highway Design Process.
Part 18: Regression Modeling 18-1/44 Statistics and Data Analysis Professor William Greene Stern School of Business IOMS Department Department of Economics.
Analysis of Individual Variables Descriptive – –Measures of Central Tendency Mean – Average score of distribution (1 st moment) Median – Middle score (50.
Business Statistics - QBM117 Statistical inference for regression.
Correlation and Regression Analysis
Simple Linear Regression Analysis
Relationships Among Variables
12 Autocorrelation Serial Correlation exists when errors are correlated across periods -One source of serial correlation is misspecification of the model.
Correlation & Regression
Linear Regression.
Regression and Correlation Methods Judy Zhong Ph.D.
LEARNING PROGRAMME Hypothesis testing Intermediate Training in Quantitative Analysis Bangkok November 2007.
Chapter 6 & 7 Linear Regression & Correlation
The Empirical Bayes Method for Safety Estimation Doug Harwood MRIGlobal Kansas City, MO.
Network Screening 1 Module 3 Safety Analysis in a Data-limited, Local Agency Environment: July 22, Boise, Idaho.
2-1 LOW COST SAFETY IMPROVEMENTS The Tools – Identification of High Crash Locations – Session #2.
Evaluation of Alternative Methods for Identifying High Collision Concentration Locations Raghavan Srinivasan 1 Craig Lyon 2 Bhagwant Persaud 2 Carol Martell.
1 CEE 763 Fall 2011 Topic 1 – Fundamentals CEE 763.
Creating a Critical Crash Rate Report Using Existing Tools and Available Information Heath Hoftiezer, P.E., P.T.O.E Civil Engineer/ P.E. (Traffic) City.
The Examination of Residuals. Examination of Residuals The fitting of models to data is done using an iterative approach. The first step is to fit a simple.
Ch4 Describing Relationships Between Variables. Section 4.1: Fitting a Line by Least Squares Often we want to fit a straight line to data. For example.
SPFs Applications by State DOTs John Milton Ph.D., P.E., Washington State Department of Transportation National Safety Performance Function Summit July.
Role of SPFs in SafetyAnalyst Ray Krammes Federal Highway Administration.
Chapter 4 Linear Regression 1. Introduction Managerial decisions are often based on the relationship between two or more variables. For example, after.
© Copyright McGraw-Hill Correlation and Regression CHAPTER 10.
Chapter 16 Data Analysis: Testing for Associations.
University of Minnesota Intersection Decision Support Research - Results of Crash Analysis University of Minnesota Intersection Decision Support Research.
Business Statistics: A Decision-Making Approach, 6e © 2005 Prentice-Hall, Inc. Chap 13-1 Introduction to Regression Analysis Regression analysis is used.
1 Tobit Analysis of Vehicle Accident Rates on Interstate Highways Panagiotis Ch. Anastasopoulos, Andrew Tarko, and Fred Mannering.
Putting Together a Safety Program Kevin J. Haas, P.E.—Traffic Investigations Engineer Oregon Department of Transportation Traffic—Roadway Section (Salem,
CE 552 Week 9 Crash statistical approaches Identification of problem areas - High crash locations.
July 29 and 30, 2009 SPF Development in Illinois Yanfeng Ouyang Department of Civil & Environmental Engineering University of Illinois at Urbana-Champaign.
Continuous Risk Profile: A Simple Method for Identifying Sites for Safety Investigation. Koohong Chung, Ph.D. California Department of Transportation Highway.
Copyright © 2011 by The McGraw-Hill Companies, Inc. All rights reserved. McGraw-Hill/Irwin Simple Linear Regression Analysis Chapter 13.
Fall  Crashes are “independent” and “random” events (probabilistic events)  Estimate a relationship between crashes and covariates (or explanatory.
Variance Stabilizing Transformations. Variance is Related to Mean Usual Assumption in ANOVA and Regression is that the variance of each observation is.
Session 2 History How did SPF come into being and why is it here to stay? Geni Bahar, P.E. NAVIGATS Inc.
Lecturer: Ing. Martina Hanová, PhD.. Regression analysis Regression analysis is a tool for analyzing relationships between financial variables:  Identify.
Evaluating the performance of three different network screening methods for detecting high collision concentration locations using empirical data Prepared.
LOW COST SAFETY IMPROVEMENTS Practitioner Workshop The Tools – Identification of High Crash Locations – Session #2.
DETECTION AND ASSESSMENT OF SAFETY PROBLEMS WITHIN ROAD TRANSPORT DECISION MAKING Prof. Dr. Nikolay Georgiev eng. Violina Velyova ‘Todor Kableshkov’ University.
Caldwell and Wilson (1999) 1. Determine primary rating factor for a road section based on traffic volume and user types 2. Primary rating factor is then.
AP Statistics Chapter 14 Section 1.
Kin 304 Regression Linear Regression Least Sum of Squares
Chapter 12: Regression Diagnostics
BPK 304W Regression Linear Regression Least Sum of Squares
BIVARIATE REGRESSION AND CORRELATION
Exploratory Analysis of Crash Data
Before-After Studies Part I
26th CARSP Conference, Halifax, June 5-8, 2016
Doug Harwood Midwest Research Institute
Checking Assumptions Primary Assumptions Secondary Assumptions
Nazmus Saquib, PhD Head of Research Sulaiman AlRajhi Colleges
Canadian Associate of Road Safety Professionals Conference May 2019
Presentation transcript:

Incorporating Temporal Effect into Crash Safety Performance Functions Wen Cheng, Ph.D., P.E., PTOE Civil Engineering Department Cal Poly Pomona

Presentation Outline General safety background Description of the research method and crash data Illustration of the results Discussion and Conclusions

General Safety Background

Crashes in Real Life: Huge Burden

Crashes in the U.S. 42,116 fatalities (i.e., 115 persons killed/day) 1.51 fatalities/100M VMT fatalities/100K Population 231 Billion Economic Cost 41% alcohol-related fatalities 29.7% Speed-related fatal crashes Source: 2005 NHTSA

Crashes in California Source: 2005 CA Statistical Abstract

Powerful Tool: SPF Definition: Based on AASHTO HSM, SPFs are equations that estimate expected average crash frequency as a function of traffic volume and/or roadway characteristics. It is a powerful tool to determine the influential factors of crashes. Popular Modeling Technique: The basic negative binomial regression which can account for the over- dispersion of crash data. – Main Drawback: the temporal effect associated with cross section and time series data is deficiently considered.

Hence, the main research objective Conduct Generalized Estimating Equations (GEEs) which can provide an extension of generalized linear models to the analysis of longitudinal data and account for the correlation in the repeated observations for a given intersection.

Research Method and Data Descption

Normal Linear Regression Requires Three strong assumptions Normally distributed errors (i.e., residues) Constant variance of errors No relationships among the independent variables (i.e., regressor variables, or predictors)

In This Study The independent variable y (accident number) is nonnegative count data, which is not normally distributed Therefore, the Normal Linear Regression is not appropriate herein. Instead, we can use a Count Data Regression Model.

The Typical Negative Binomial Model has advantage over Poisson Model in accounting for the over-dispersion of crash data. However, the temporal effect of the crash data over the time period is not considered

Two GEE Models in this study The GEE model with autoregressive correlation It weighs the correlation between two observations by their separate gap (order of measure). As the distance increases the correlation decreases. The GEE model with unstructured correlation structure It assumes different correlations between any two observations taken at the same time. Both Models sufficiently consider the temporal effect associated with cross section and time series data!

Crash Data Used Obtained through Crossroads Collision Database 298 intersections from the City of Corona, CA – 141 signalized – 157 unsignalized Total crashes: 7594 Crash period: 10 years (2000~ 2009)

Data Used in the Study Major road speed limit Minor road speed limit Major road ADT Major road ADT year Minor road ADT Minor road ADT year Signalized intersection (Yes or No) Total number of through lanes Presence of at least one exclusive right turn lane on major road (Yes or No) Presence of at least one exclusive left turn lane on major road (Yes or No) Presence of at least one exclusive right turn lane on minor road (Yes or No) Presence of at least one exclusive left turn lane on minor road (Yes or No)

Research Results

Modeling Results of the Three Models

Discussion of Results GEE models have slightly higher estimated standard errors than do traditional negative Binomial models – reason: accounting for the temporal correlation will inflate the standard errors. The three models produce unequal coefficients, which means considering temporal effect of crash data could result in different modeling estimations.

Discussion (Cont’d) The crash data from other states or places might yield different results. The impact of the modeling estimation difference (due to the temporal effect) on other safety applications (e.g. hot spot identification) needs to be further evaluated.

THANK YOU!