Data Acquisition

Firstly, I acquired the data for the dissemination areas from Statistics Canada as it would be most effective to study the city by its smallest survey area. The shapefile was then filtered by selecting the attributes relating to Vancouver. From there a new layer was created and exported to the geodatabase from those selected attributes.

Secondly, I acquired the 2016 census data from CHASS to conduct the food desert analysis. For the hotspot analysis, I decided to download the data for household total income groups in 2015 for private households for incomes. The incomes range from under $5,000 to $100,000 and over. For the SCMC, I decided to download the data for employment rate, single status (Not married and not living common law), lone-parent families and couple census families with children.

For the grocery stores, I acquired the shapefile from Open Data Vancouver Portal. The original file contained all the business licenses so I filtered the table by the current year and by the food retail category. Lastly, I obtained the transit routes and stop shapefiles from the Abacus data portal.

Methods

#1) Hot Spot Analysis

The shapefile for Vancouver’s dissemination areas is first joined with the CSV containing data for the household income. In the table, I created a new field called average in order to store the values for the average for each dissemination area. The average was calculated using the summarize fields tool to find the mean of the 17 columns in the CSV. A hotspot analysis was then conducted to locate clusters of high income and low income areas.

#2) Distance Analysis

A 1 km buffer was used to display the accessibility of each grocery store. 1km was chosen as it is a reasonable walking distance. Transit routes and stops were also included into the analysis to show that places not within reach of a grocery may also use public transit to access the stores, although this will come at an additional price due to the transit fares.

#3) Spatially Constrained Multivariate Clustering

From the Vancouver DA shapefile, another join was made to the CSV containing the data for employment rate and family characteristics. A SCMC analysis was done with the parameters splitting the clusters into four areas. The buffer layer was also included in the final output to show the population characteristics most affected by the food deserts.