Traffic engineering & control · 1976 · 12 citations · 0 references
Natural SciencesPhysicsData VisualizationFlow VisualizationFluid DynamicsFlow MeasurementStatistical GraphicsSpeed-flow RelationshipsStatistical Significance Testing
The author examines the use of speed/flow and speed/concentration plots. Often the speed/flow data show a large amount of scatter and the corresponding speed/concentration plots show a marked tendency to which a downward straight line may be fitted. This straight line can be reconverted to a parabola on the speed/flow diagram, establishing a significant relationship between the two variables which are totally uncorrelated giving a false measure of capacity. Normally distributed random values of speed are used to illustrate the abuse of the relation. A further danger in the transformation process is that several rectangular hyperbola give an equally good fit to the speed/concentration data as the straight line; it is shown that these produce significant changes in the shape of the reconstructed parabola on the speed/flow diagram. The shape and statistical significance of a speed/flow relation should not be inferred from a speed/concentration relation unless the latter is very clearly defined over a range of values. /TRRL/