Grouping Algorithm for Statewide Loadmeter Surveys

A multivariate statistical procedure consisting of modified regionalization and discriminant analysis was developed to identify loadometer stations collecting redundant information. The statistical technique represents a methodology for determining station similarities through the formulation of homogenous groups. Thus, each loadometer station is assigned to a group, but only one group. Traffic composition was used as the measure of station identification. Based upon previous studies relating classification counts to truck axle-weight, it was assumed that the percentage of trucks in six different classes found in the traffic stream was sufficiently representative of actual truck weight for use in loadometer surveys. Application of the statistical procedure to 17 West Virginia loadometer stations for 1963, 1965, 1966, and 1967 identified three groups. Six stations in one group were identified as being stable over time so information is already being obtained at the other stations. Other stations in the other two groups also tended to show association.