Assessing Speed Camera Effects on Collisions in Toronto

Link to project: https://ryerson.maps.arcgis.com/apps/dashboards/e09127998c21447ea85ce713c1502fe4
Author: Steven (Shucheng) Wang
Geovis Project Assignment @RyersonGeo, SA8905, Fall 2021

Introduction

In 2017, 50 Automatic Speed Enforcement (ASE) cameras were installed throughout Toronto. These cameras work by taking pictures of vehicles which are speeding, and then issuing a fine to the owner of the vehicle. 2 Cameras are allocated to each ward located mainly near school zones for a total of 50, which will eventually be rotated out for a different set of 50 in different locations. This strategy is meant to reduce collisions by having people slow down in areas where the ASE cameras are present.

Figure 1: ASE camera in Toronto

To visualize whether the installation of these cameras has made a difference on collisions in Toronto, I decided to use ArcGIS Dashboards. ArcGIS Dashboards is a tool that presents spatial data and associated statistics in an interactive format, allowing the user to get the answers to questions that they want.

In order to put this together, I collected collision data from the Toronto Police Public Safety Data Portal which includes data on collisions throughout Toronto from 2006 until 2021. I also collected data on the ASE locations from the Ontario Open Data Portal, and opened both datasets in ArcGIS Online to edit their symbology before adding it to a dashboard.

FIgure 2: Preparing the data for use in the dashboard

Now that the map was ready, I started to configure the actual dashboard. The main elements that I considered essential to include were:
• A filter system, to allow users to filter collisions under certain conditions
• A pie chart, to allow users to visualize the percentage of each type of collision depending on their filters
• A line or bar graph to allow users to see the distribution of collisions temporally.
These can all be easily added to a blank dashboard and configured using the “+” button on the ArcGIS Dashboards top header. The final dashboard with all the previously mentioned elements and the map frame can be seen below:

Figure 3: Complete dashboard

Dashboard Elements and Functions

The first element we will be looking at is the side panel on the left, which contains the date selector as well as multiple category selectors for different attributes. Each one opens an accordion-style menu when clicked, displaying all available filters for that particular category. These filters can be toggled on or off, and the map frame in the centre will reflect any filters made.

Figure 4: Visibility selector with rain toggled

The next element is the serial chart at the bottom of the dashboard, which contains two graphs stacked on top of each other. The first one is a line graph of the collisions by date all the way from 2006 until now, and the second one is a histogram of the collisions per hour based on a 24-hour clock. Both graphs contain a time slider at the top which can be used to zoom in and look at a particular time period in detail. However, the time slider is purely for viewing purposes and will not affect the map.

Specific time periods can also be selected by clicking on the graph and dragging your mouse over them, or by holding CTRL and clicking the time periods as well. For example, if you only wanted to see collisions from 2017 and onwards, you could click and drag your mouse over the part of the line graph where 2017 starts all the way until the far right side. Unlike the time slider, selecting time periods this way will reflect on the map frame.

Figure 5: Serial graphs

The final elements are the legend and the pie chart to the right of the map frame. The legend displays the categorization for each data point, like which ASE camera is currently active vs. planned, or which collisions resulted in fatalities vs. injuries. The pie chart is stacked on top of the legend and displays the distribution of collision type. Similar to the serial charts, the pie chart will adjust to fit the map extent and the filters chosen. However, the legend is static and will not change regardless of filters or map extent.

Limitations & Conclusion

While a dashboard like this can be convenient in many ways, there are some limitations. For example, for the serial graphs there is no indication in the UI that time periods can be selected at all. I only found out about the function when I accidentally clicked on it; before, I had assumed that the time slider would provide that function and was confused why the data points on the map did not change when I adjusted the time slider. Additionally, it is much more difficult to see when time periods have been selected on light mode than on dark mode, which is why I set this dashboard to dark mode.

Another limitation is that there is no real way to conduct spatial analysis beyond the functions outlined earlier. Common tools like creating buffers or finding intersections that would be present in ArcMap/ArcGIS Pro/QGIS are nowhere to be found. You could do these analyses in those programs, create a layer from said analyses, and then import it into a webmap as a layer before adding it to a dashboard, but that would require you to rework the entire dashboard.

Overall, dashboards are a convenient way of allowing users who aren’t familiar with GIS to manipulate and visualize spatial data. It can be a great way to simplify data and create a neat tool that can identify trends or statistics at a glance. However, it is important to note that due to its limitations, its utility will depend greatly on your use case.