Section I

Selection of an Appropriate 'statistical Model' for Baseline Road for Low Volume Traffic Flow

Authors:

K. S. Weerasekera
Email K. S. Weerasekera

Abstract

It was attempted to select an appropriate'statistical model' for a set of observed traffic data on Baseline road, Colombo. A traffic counting survey was conducted to collect traffic data along Baseline road on a weekend. A Sunday afternoon was selected to conduct the traffic survey to ensure an uninterrupted flow condition without congestion at the site. Data obtained from the survey was analysed and appropriate counting interval and time period for the development of the model were selected.

A counting interval of 10 seconds was conducted, and later a spectrum of 10,20 & 30 second intervals were computed and plotted. It was,observed that 10 second counting interval gave the best shape graph with a maximum frequency at one vehicle per 10 second interval and a positive skew. The observed frequency distribution for L0 second interval, showed a reasonably well distribution with a single mode, and a positive skew tailing to the right. The 20 and 30 second intervals showed a deviation from the regular pattern further and further. Hence the pattern seems to deteriorate with the increasing time interval.

The analysis proved that the sample could be identified as Negative Binomially distributed. A goodness-of-fit test was used to check the difference between field data and expected frequency. Negative Binomial Model (with k=2) gave the smallest Chi-square value with parameters of p = 0.38, which best fitted to the obtained sample data. Hence it could be concluded that Baseline road traffic was 'randomly' distributed under uninterrupted flow conditions, and Negative Binomial Model was the statistical model that best described the traffic flow conditions at the site at the time.

Keywords:

TrafficFlowModelling NegativeBinomiallyDistribution

Published on 30 Jan 2010

Peer Reviewed