1 CIVL7415: Transport Data Analytics School of Civil Engineering Semester 1, 2022 – Project 2 The University of Queensland To be submitted through BlackBoard Due: 4:00pm Friday 29 April 2022 100 Marks total; points for each question as marked Within the Queensland Department of Transport and Main Roads, Translink is the division responsible for the funding and oversight of all public transport services in Queensland. For this project, we will be exploring the available public transport services, and the demand for these services, across South East Queensland (SEQ). For the purposes of this project, we define SEQ using the local government areas as far west as Toowoomba, as far south as the Gold Coast, and as far north as the Sunshine Coast. In this project we will examine the distribution of public transport services as they vary across space in South East Queensland. The critical question for this project is: After accounting for distance from the Brisbane CBD, are there spatial patterns in the level of access provided to public transport Specifically, is public transport access “biased” towards certain income or housing groups, or towards Indigenous Australians To measure “access”, we will use the “Public Transport Accessibility Level”, or PTAL. The PTAL metric was originally developed by Transport for London to capture the average time spent walking to, and waiting for, bus and train service in greater London. The PTAL metric captures this accessibility at a fairly fine level of spatial detail. A report on the method used to calculate PTAL is included on Blackboard with this project. To measure the PTAL, we also need a number of important metrics on public transport service. We will use the locations of bus and train stops, and the timetables at those stops, to measure public transport service. From the General Transit Feed Specification (GTFS), the Blackboard site contains the files “stops.txt” and “stop_times.txt”, which provide information on the spatial location of stops and the timetable at each stop. Geographies (shapefiles) for Mesh Blocks and for Spatial Area 1 (SA1) as defined by the Australian Bureau of Statistics (ABS) are included in the data online; these have already been processed to include only those shapes within the SEQ region. Finally, selected data from the Australian Bureau of Statistics (ABS) at the level of Spatial Area 1 (SA1) for Queensland is also given. This takes the form of Table G02 and Table G07, recorded at the SA1 level. Descriptive information on these two tables is show at the end of this project statement. A short summary of the data files available in Blackboard is provided on the next page. Task 1 (10 marks) Calculate the number of public transport stop-times for every stop, during the morning peak period from 6:30 am to 9:00 am. You can do this in R or whatever means you choose. [Note that the file “stop_times.txt” is too big to open in Excel.] Submit this result as a CSV file, with the stop id, stop name, and number of stop-times, with your project report. Task 2 (40 marks) Calculate the PTAL for all SA1 areas in SEQ. To do this, estimate the average distance from a Mesh Block Centroid to the nearest public transport stop. To convert this to walk time, assume an 2 average walking speed of 4.8 km/hr. Then, estimate the waiting time at that stop using the method described in the PTAL report, using the stop-times derived in Task 1 and the method described in the PTAL method. This would also then lead to an “Equivalent Doorstop Frequency”, or EDF, as described in the PTAL method. The net result should be an estimate of the PTAL value for each Mesh Block. [Do not convert the value into an “Index”.] Use a simple average of
these values at the Mesh Block to create an average PTAL value for each SA1. Plot these PTAL values on a map for SEQ, at the level of the SA1. Also, submit this result as a CSV file, with the SA1 code and the PTAL value. Task 3 (25 marks) Conduct a linear regression of the PTAL value as a function of distance from the Brisbane CBD. As the CBD, use SA1_7DIGIT = 3110509 or SA1_MAIN16 = 30501110509. [You may also consider transformations of the “distance” variable, e.g. distance squared, square root of distance, etc.] From this regression, determine if there is spatial correlation among the residuals, by checking both Moran’s I for SEQ and the Getis-Ord Gi* for each SA1 area. Include the output of the regression, the value of Moran’s I, a map of SEQ with the values of the Getis-Ord Gi* for each SA1 area, and a CSV file with the SA1 code and the Getis-Ord Gi*. Also include a 1-page interpretation of the output values in the regression, of Moran’s I, and of the Getis-Ord Gi*. Task 4 (25 marks) Repeat the analysis from Task 3, but feel free to include other ABS statistics associated with the SA1. These could include any statistics from Tables G02 or G07, as provided, or any other SA1 statistics found from the ABS website. From this regression, determine if there is spatial correlation among the residuals, by checking both Moran’s I for SEQ and the Getis-Ord Gi* for each SA1 area. In addition, if you find spatial correlation, conduct a Geographically Weighted Regression (GWR) using the same variables. In your output, include the output of the regression, the value of Moran’s I, a map of SEQ with the values of the Getis-Ord Gi* for each SA1 area, and a CSV file with the SA1 code and the Getis-Ord Gi*. Also include a 1-page interpretation of the output values in the regression, of Moran’s I, and of the Getis-Ord Gi*. Data table on next page. 3 Data provided File name Relevant contents SEQ MB region.shp The geography (shape file) for Mesh Blocks located in South East Queensland. The mesh block is the smallest geography for data aggregation in the Census conducted by the Australian Bureau of Statistics (ABS). SEQ SA1 region.shp The geography (shape file) for Statistical Area 1 zones located in South East Queensland. The SA1 aggregates mesh blocks and provides the lowest level of detail of population statistics collected in the Census by the ABS. 2016Census_G02_QLD_SA1.csv From the ABS data of the 2016 Census, Table G02 has median values of income by person, by family, and by household, as well as median rent, median mortgage, and average household size. These data apply at the SA1 level. 2016Census_G07_QLD_SA1.csv From the ABS data of the 2016 Census, Table G07 has the total number of persons of Indigenous and non-Indigenous heritage living in an SA1, by age and gender. stops.txt This comma-delimited file has the identification number, name, and location (latitude and longitude) of every stop in Translink’s network for South East Queensland. This file is taken from the General Transit Feed Specification (GTFS) for South East Queensland. stop_times.txt This comma-delimited file has the information from the timetable for every bus, train, and ferry in South East Queensland. The primary identifiers are the route, the service (scheduled trip by a bus, train or ferry), the stop, the stop identifier, and the scheduled time of arrival and departure at the given stop for that service. Also, the service is identified by weekday, Saturday, and Sunday. 13-06-2013.csv This file contains the boarding and alighting stop of every public transport trip made in South East Queensland using a go card, on 13 June 2013 (a Thursday). A separate sheet indicates the data items in this file.