In early 2015, Nicolas Kruchten collaborated with a reporter from the Montreal Gazette, Roberto Rocha, to analyze a dataset containing information about 1.4 million service requests received by the City of Montreal from its citizens. The resulting article was entitled "Montreal's 311 records shed light on residents' concerns — to a point" and credits Nicolas at the bottom. The dataset, obtained from the city's Gestion des demandes clients (GDC) system via an Access to Information request and also worked on by Stéphane Guidoin, covered the five years from 2008 to 2012 and contained only the date and a very short description for each request, and in most cases, an address. Notably missing was any indication of resolution time or correspondance with the originating citizen. The service requests were received by the city through its 311 phone line or at service counters throughout the city.
In these pages, you will be able to walk through the process of analyzing the 1.2 million requests which were linked to a specific borough through an interactive pivot table such as the one you see below. As you advance through the steps of the analysis, the table will reconfigure itself to show you different views of the data. You can interact with the pivot table at any time by dragging the fields around to modify the tables and graphs, or by hiding/showing information by clicking on the little triangles.
Let's get started!
The original dataset contained more than 4,000 unique request descriptions. The pivot table below is loaded with a partially-processed dataset: the requests have been grouped into 18 broad categories. You can find examples of the various descriptions, as well as a detailed explanation of the categorization process here. The chart below shows the distribution of requests by category; the categories are ordered by overall request volume.
On an administrative level, Montreal is divided up into 19 boroughs, some of which were merged into the city in 2002 and then reorganized in 2006. Below is the distribution of requests by borough and category. The boroughs are ordered by population. You can mouse over the bars to see which category is which and how many requests it represents within each borough. Click on the map for a handy reference.
In the table below, we are looking at the percentage of requests that each category represents within each borough. The redder the cell, the more that category preoccupied that borough.
In the table below, we are looking at the percentage of requests that each year represented within a category. The redder the cell, the more requests in that category occurred in that year.
In the table below, we are looking at the percentage of requests that each month represented within a category. The redder the cell, the more requests in that category occurred in that month.
The Security category stood out in both time and space (a large percentage of the requests in Saint-Laurent and L’Île-Bizard–Sainte-Geneviève, and as being much higher in 2008/2009 than in 2011/2012), so let's take a closer look. In the graph below we have Security requests by borough over time by month. We see a dramatic pattern: almost the entire change in request volume over the years is due to Saint-Laurent. Looking closely, L’Île-Bizard–Sainte-Geneviève also displays a cyclical pattern with more requests in summer. You can click on Saint-Laurent in the legend to hide that borough's line and see the others more clearly.
The Parking category stood out as being much higher in 2008/2009 than in 2011/2012. Taking a closer look at just that category we can pin-point the change in May/June 2010, and we can see that unlike the Security category, it is not a case of a single anomalous borough, but rather spread out across many boroughs.
The Selective Collection category stood out as spiking up in 2012 and being a big percentage of the requests in Saint-Laurent. Taking a closer look, we can see that Saint-Laurent figures prominently, but in small spikes throughout the period. Additionally, we see that the rise in 2012 occurs across all boroughs starting in April/May.
The Bulky Collection category stood out as being a huge percentage of the requests in Rivière-des-Prairies–Pointe-aux-Trembles and as spiking in 2008. Taking a closer look, RDP-PAT dominates throughout the 5-year period, but we also see the smaller pattern of every borough having more requests about this in 2008. If you look closely, you will see that there are also regular spikes in this category of requests in May and September in Lachine
The Taxes category stood out as spiking in February, and as being a big deal in LaSalle and Saint-Léonard. Taking a closer look, we see that most boroughs have a spike in February, but that Saint-Laurent, Saint-Léonard and LaSalle have a different pattern that spikes in April and tails off more slowly.
The Environment category stood out as spiking in the month of May. Taking a closer look, we can immediately see that the May spike is solely due to Saint-Laurent. There is also a smaller pattern of spikes in March in Le Sud-Ouest, and some spikes in Ville-Marie, Rosemont–La Petite-Patrie and Pierrefonds-Roxboro.
This analysis began with an overview chart of requests volumes by category (replicated below) and then proceeded to look at places and time periods where request volumes deviated from this overall pattern. Most of the deviations turned out to be attributable to seasonal effects or borough-specific policies or differences in the way in which different boroughs use the Gestion des demandes clients (GDC) system. In the aggregate, however, most of the time and in most places, the distribution of requests does follow the overall pattern, so we can look at the chart below and conclude that it more or less sums up the bulk of the service requests in the GDC system. That said, there may be more subtle patterns and stories of interest in the "Other" category, which represents 6.5% of the requests.
Beyond looking at types of requests over time and space, it would certainly be interesting to examine the outcomes of service requests, for example what was done in response to a request and how long it took or how much it cost, but unfortunately this information was not present in the dataset obtained by the Gazette.
Questions, feedback or comments on this analysis are welcome by email at email@example.com!
This is a data story by Nicolas Kruchten using PivotTable.js