Analysis of inbound freight to Newfoundland

• Main Author: Batstone, Paul S.
• Format: Thesis
• Language: English
• Published: Memorial University of Newfoundland 1990
• Subjects: Newfoundland
• Online Access: https://research.library.mun.ca/5311/; https://research.library.mun.ca/5311/1/Batstone_PaulSelby.pdf; https://research.library.mun.ca/5311/2/Batstone_PaulSelby.pdf

The prime purpose of this research was to analyze the historical patterns of inbound general freight to Newfoundland by mode of transport, to identify the socio-economic factors which chiefly influence the inbound traffic, and to develop and test a forecasting model. A secondary objective was to investigate and test several non-causal forecasting techniques whose results could be compared with those of the model tested. -- The approach to this transportation planning topic was guided by the well developed method of analysis used in traffic engineering and urban transportation planning. Having reviewed the literature and collected a suitable data set, a preliminary statistical analysis was conducted. This was followed by the application of linear regression with the appropriate testing of the underlying assumptions using the t-test, the F-test and the Bartlett's test. Other methods used included the calculation of simple ratios of traffic volume to the various socio-economic factors, and time series and smoothing techniques. -- The historical pattern of general freight to Newfoundland has demonstrated constant growth over the last 20 years with considerable variability in the split between the rail, truck and direct shipping modes. Current inbound traffic by all modes is in the order of 1 million tonnes annually. Using multiple linear regression, the combination of population and the value of retail trade was found to influence the inbound traffic to the greatest extent. Simple ratios of inbound traffic to selected socio-economic employment ratio demonstrated the least difference between forecast and actual value. -- The best straight line fit was found to be the combination of population and the value of retail trade with the use of regression analysis. The resulting equation had an R-Square of 80%, which is considered an acceptable degree of accuracy for the industry, and intuitively made sense. The predictions calculated from non-causal techniques resulted in similar orders of accuracy. -- In a broader vein, ...