The new map can be found here. A screenshot showing southern Durham is at left.
What patterns do you see in the data? Let us know in the comments.
The new map can be found here. A screenshot showing southern Durham is at left.
What patterns do you see in the data? Let us know in the comments.
 |
| IAN WILLMS/TORONTO STAR |
Regular readers of this blog will have seen several iterations of words to the effect that a map is being produced because improved tools have made it possible.
When Statistics Canada published a file last year of 2006 crime rates for Toronto broken down by 531 census tracts, it wasn't possible to map it on line, and I more or less forgot about it until this week until I found it in an ideas spreadsheet. (Yes, I have an ideas spreadsheet - doesn't everybody?)
The data is from 2006, but very finely textured. I can only imagine the work involved. The authors explain:
Incidents that were located on the boundaries of many census tracts were divided in that number of census tracts. For example, an incident located on the boundary of 3 census tracts was counted as 0,333 incidents in each of the 3 census tracts.
As far as I can tell, the researchers ended up mapping 26,022 violent crimes. Property crimes were dealt with separately, and will be the subject of a different map.
Also, there was an attempt to relate crime rates to both residents and visitors in an area. One problem with our neighbourhood-based crime maps is that they divide the number of reported crimes in a neighbourhood by the number of residents. I suspect that this overstates the crime rate as experienced by residents in neighbourhoods with many outsiders, like those downtown. It's never been clear to me what to do about the problem other than to flag it for the reader.
Here is a top 20 list:
| 1 | 5350032 | Carlton/Dundas/Jarvis/Parliament |
| 2 | 5350312.05 | SE of Jane/Finch |
| 3 | 5350033 | Queen/Dundas/Jarvis/Parliament |
| 4 | 5350316.05 | NE of Jane/Finch |
| 5 | 5350207 | Off Kipling N of Lake Shore |
| 6 | 5350039 | Queen/Dundas/Spadina/Bathurst |
| 7 | 5350080.02 | Danforth W of Victoria Park to Main |
| 8 | 5350003 | SW Parkdale |
| 9 | 5350248.02 | Martin Grove/Albion Road/Kipling/West Humber |
| 10 | 5350312.02 | SE of Driftwood/Finch |
| 11 | 5350292 | Sheppard/Highbury/Jane/401 |
| 12 | 5350018 | DVP/Eastern/Carlaw/Queen |
| 13 | 5350031 | Gerrard/DVP/Dundas/Parliament |
| 14 | 5350173 | NE of Jane/Eglinton |
| 15 | 5350358.03 | Along Kingston Rd SE of Lawrence |
| 16 | 5350312.04 | SW of Jane/Finch |
| 17 | 5350341.03 | NE of Danforth/Victoria Park |
| 18 | 5350069 | Bain/Pape/DVP/Gerrard (includes Don Jail) |
| 19 | 5350097.01 | NW of Bloor/Dufferin |
| 20 | 5350337 | SW of Midland/Kingston Rd |
In an e-mail back-and-forth this week, Statistics Canada explained that the violent crimes tracked were:
Homicides: murders, manslaughters, infanticide
Attempted murders
Assaults
Sexual assaults, sexual offences (invitation to sex. touching, luring via computer, etc.)
Robberies
Criminal negligence causing bodily harm
Firearm offences: use of, discharge, pointing
Forcible confinements, kidnappings, abductions
Criminal harassment, uttering threats, threatening or harassing phone calls
Other violent violations: trafficking in persons, arson causing disregard for human life, intimidation, hostage-taking, incest, voyeurism, etc.
 |
| COLIN PERKEL/CANADIAN PRESS |
As with several recent maps, this is an exercise in re-presenting data with better tools than were available when the original map was first produced.
The map shows two strong patterns:
The FOI request in this case raised an emerging issue with how the federal government handles requests for electronic information. In this case, I filed a request asking for an electronic file showing the first three characters of the postal codes of 2007 recruits, with the reserves and regular force in distinct records.
This took some time to appear, and the reason became clear when DND released a CD-ROM with several .pdf files containing scanned lists containing the postal code information for each individual recruit - 203 pages for the regular force and 153 for the reserves. The tables in these lists, which are images, can't be saved as tables in Windows. The only way of getting a total for any given FSA is to print out the .pdf and count them manually. For Ontario, this meant going through 77 pages of printouts with 37 records per page, counting up totals for 500-odd FSAs. Bear in mind that this started life as an electronic record. I'd produce a military recruiting map for the whole country if I could make time for the manual counting process involved.
I wrote this off as a quirk of military culture until I dealt recently with another federal department which released postal code-based data by mailing a 40-page printout of all 1600-odd FSAs in the country, forcing me to manually enter the values related to them in Excel. (I'm taking a break after reaching the Manitoba-Saskatchewan border.)
The issue here is that data reduced to paper form loses much of its usefulness. The effect is to take power away from the recipient of the data (and by extension in this case from you as a citizen) and conserve it in a government institution as much as possible. Unless the user is bloody-minded enough to re-enter it manually, which of course is only possible at a certain scale.
Halifax-based journalism professor Fred Vallance-Jones blogged about this issue recently:
In late 2007, the Treasury Board Secretariat, which oversees access to information in the federal government, advised departments to "cease responding to ATIP requests in electronic format until they are certain that any potential risks (of inadvertently releasing severed information) have been addressed."
This doesn't have to be a concern. If there is a fear about information being hidden in a file - and there are certainly examples of hidden data being extracted from censored files - then all the agency has to do is save the releasable information as a text file, where there is no place for data to hide.
Ontario ministries, by contrast, will usually release a spreadsheet on CD-ROM - this is an issue with the federal government specifically.
 |
| RICHARD LAUTENS/TORONTO STAR |
As the map shows, the deaths have been distributed between the central city, the inner suburbs and the 905. Do you see a pattern? Let us know in the comments.
 |
| CANADIAN PRESS |
Map: Average annual drug charges per 1,000 residents, 2004-2008
This week continues our occasional series of neighbourhood crime maps with a look at drug charges.
The most obvious thing about the map is the higher rates of drug charges more or less downtown, roughly in an area bounded by Queen/King, Bathurst, Bloor and Parliament, more or less. The missing piece of information is how many of these charges were laid against residents of any given neighbourhood, and how many against people from outside.
Looking more broadly at the city, we see the familiar check-mark shape extending diagonally down though the west end, through the downtown and up again through Scarborough. This appears on many other maps, including this one and this one.
The rate is averaged over five years because of significant year-to year fluctuation in many areas.
The map shows five years of data for all 140 neighbourhoods, as seen below:
 |
This week, we continue our crime series using neighbourhood-based crime statistics produced by the Toronto police. These use the City of Toronto neighbourhood boundaries, not the set produced by our neighbourhood project earlier in the year.
Vehicle theft is strongly concentrated in the northwest of the city, a pattern we saw more crudely in a division-based map last year. A four-neighbourhood clump centres more or less on the De Havilland plant. West Humber-Clairville, in the city’s northwest corner, has the highest rate, one of only two neighbourhoods with a rate above 10 2008 vehicle thefts per 1,000. (Yorkdale-Glen Park is the other. )
Here are the top 10:
I took all five years worth of data to get something recognizable as a neighbourhood-scale homicide rate.
High homicide rates (by Toronto standards) concentrate in four areas: Downsview/Rexdale/Jane-Finch; Parkdale and Niagara; the east downtown; and northwestern Scarborough.
Here are the top 10:
Individual homicides (omitting 2004 and including 2009 to date) can be seen here.
The maps show five years of data for all 140 neighbourhoods, as seen below:
 |
| SHUTTERSTOCK |
This week, we begin an occasional series on crime, using neighbourhood-based crime statistics produced by the Toronto police. These use the City of Toronto neighbourhood boundaries, not the set produced by our neighbourhood project earlier in the year.
Before I saw the map, I expected more burglary in high-income neighbourhoods, based on the idea that wealthy people would be more likely to have things worth stealing. That turns out not to be consistently true, though the Bridle Path and Casa Loma do have high rates. The stronger relationship seems to be with areas around universities, with U of T, York and (almost) Ryerson in areas with very high burglary rates.
Here are the top 20:
1 University
2 Bay Street Corridor
3 Kensington-Chinatown
4 Moss Park
5 Bridle Path-Sunnybrook-York Mills
6 East End-Danforth
7 Casa Loma
8 York University Heights
9 Annex
10 Church-Yonge Corridor
11 Waterfront Communities-The Island
12 Rosedale-Moore Park
13 Woodbine Corridor
14 Clairlea-Birchmount
15 West Humber-Clairville
16 South Riverdale
17 Little Portugal
18 Roncesvalles
19 The Beaches
20 Forest Hill South
It’s an odd mixture of areas, and I’d be curious to know what readers make of it.
One striking pattern is the fall in breaking and entering between 2004 and 2008 in most areas.
This trend has been noticed elsewhere in Canada, in the United States and in England. Burglary rates in the United States are somewhat more than half what they were in 1988.
Two business/economics blogs (link, link) have floated the idea that the decline in burglaries is linked to falling prices of manufactured goods – in other words, that burglars are a casualty of free trade.
The maps show five years of burglary data for all 140 neighbourhoods, as seen below:
 |
Reporter Jim Rankin has the first part of a series on neighbourhood incarceration rates in today's paper. You may remember a similar series in 2007 - the major update this year is data including the full six-digit postal codes of jail inmates, which allows for very detailed mapping.
Rankin was at first denied this information, but eventually the Information and Privacy Commissioner office ruled in his favour on appeal (the decision can be read here)
In any case, two maps go with today's story:
It bears more than a passing resemblance to our map of male homicide victims from a few weeks ago.
 |
| AARON LYNETT/TORONTO STAR |
Map: Homicides with female victims, January 2005-June 2009
We have been mapping homicides in the GTA's five regions since January, 2005, and sorting them by a number of characteristics. (Our main homicide map, with victims sorted by age, gender, year and gun/non-gun is here.)
One distinct pattern on the map is gender - homicides with male victims are much more clearly tied to geography than homicides with female victims. This week's maps separate male from female homicide victims to show the pattern more clearly.
Homicides with male victims form a sharply defined pattern, starting in Malton and continuing east through Rexdale, then down along Jane and Weston Road into the downtown. To the east, they spread out into Scarborough in a looser way, with clusters in Flemingdon Park, north of Danforth and Victoria Park, and Malvern.
Killings of women are more diffuse on the map, with the closest thing to a real pattern along Dundas, and Yonge between Dundas and Bloor.
This is a one-off map (with data current to June 13, 2009), and will not be updated. If you are looking for an updated homicide map for the GTA after that date, go here.
 |
| CHARLA JONES/TORONTO STAR |
Map: Pedestrian accidents, 2008
This week, we look at 2008’s pedestrian accidents. A total of 1,932 were reported to police last year.
Outside the downtown core, pedestrian accidents are concentrated on Eglinton between Brimley and Markham Rd., Sheppard between Kennedy Rd. and Warden, Yonge between Steeles and the 401, and Mount Pleasant south of Eglinton.
Here are the top eight intersections for pedestrian collisions:
9 BATHURST and KING
6 BLOOR and LANSDOWNE
5 CHURCH and JANE
5 COLLEGE and SPADINA
5 JANE and WILSON
5 KEELE and LAWRENCE
5 LAWRENCE and MARKHAM
5 STEELES and YONGE
For local areas, the map works best at the highest or second-highest magnification.
Nerd box:
The larger numbers involved forced a different presentation from last week’s bike collision map. This map only shows locations where two or more pedestrian accidents happened during the year, with circles of different sizes indicating the number.
I tried mapping all the accidents, but the map looked like this:
If you don't mind the visual clutter, a map showing single accidents is here.
It works best at the highest magnification and needs some patience, but does add new information. For example: Mount Pleasant between Roehampton (a block north of Eglinton) and Balliol (a block north of the cemetery) had 11 pedestrian accidents last year, or about one per 100 metres of roadway. Roncesvalles between High Park Boulevard and Queen also had more than I would have expected - eight, or one for every 112 metres of roadway.
What patterns do you see? Let us know in the comments.
(July 11: The pedestrian and cycling accident maps worked fine on their own as embedded maps, but something about displaying them both on the same page made many browsers very unstable. They’re now linked below.)
This week, we map 1,068 bicycle accidents reported to Toronto police in 2008. The data came to us from the Toronto transportation department, which analyzes accident reports they receive from the police.
Bike accidents seem to cluster in the area west of downtown, especially on College between Spadina and maybe Dufferin, and Queen between Spadina and Ossington. Dundas, King and Bloor have fewer, though Bloor between Avenue and Bathurst has more. To the east, the Bloor cluster ends abruptly at Jarvis.
(The equivalent part of the east end, which is similar in urban design, has far fewer bike accidents, though accidents do cluster on the Danforth between Pape and Broadview.)
Bay between College and Bloor also seems to have a cluster, as do major streets in the downtown area south of Dundas and north of Adelaide or so.
Time didn’t allow an overlay showing bike lanes. (Maps can be found here and here) Informally, though, accidents seem to cluster where bike lanes aren’t, and vice versa. Exceptions seem to be the College bike lane and a stretch of the Dundas St. E. bike lane west of Pape, where there is nearly one accident per block.
110 intersections had more than one accident (the great majority with two). Here are the top nine:
7 Bay and Dundas
7 College and Crawford
5 Queen and Broadview
5 Yonge and Dundas
4 Bloor and Bathurst
4 Bloor and Keele
4 Spadina and Dundas
4 Islington and the Queensway
4 King and John
Nerd box:
The large number of points created a challenge. Our normal method of loading them into an XML file created a map which was slow to load and reload at a different zoom level, and also slow to manipulate, even in Firefox. In IE it was worse.
Our alternative, loading a KML file into Google MyMaps, failed when it refused to keep loading somewhere in the mid-hundreds. MyMaps has limits, but it’s sometimes hard to see what they are other than by trial and error.
We ended up with Geocommons, the free service that the Vancouver Sun used for its parking ticket map.
In IE, it takes anywhere from 9 to 17 seconds to load, but as far as I can tell stays loaded once that has happened. The main difference, as far as I can tell, is that it uses Flash.
In any case, the look and feel of the map is different from what you may be used to seeing – let us know what you think in the comments.
The full-sized map can be seen here.
Next week, we will look at pedestrian accidents.
 |
| TARA WALTON/TORONTO STAR |
Marijuana grow op busts in Toronto were down in 2008, at 145 compared to 256 in 2007 and 254 in 2006.
However, the pattern is similar (see the 2007 map or 2006 map for comparison), with grow operations concentrated in parts of the city where low income meets low density.
We see the northern tips of the V or check-mark shape visible on other maps (for example the homicide map or map of low household incomes)
As before, grow ops are most strongly concentrated in northern Scarborough and the Hwy. 400 corridor north of Hwy. 401, which gives 31 Division and 42 Division the city's highest rates.
 |
| REUTERS |
This week, we look at military recruiting in the GTA, using 2007 enlistment data.
Reporter Allan Woods looked at recruiting a few days ago:
While the rest of the country trembles in fear of layoffs and unemployment numbers soar, the demand for soldiers, technicians and other specialists in the Canadian Forces outpaced actual growth for yet another year, the Department of National Defence said in an annual report to Parliament. The military has tempered force projections and launched massive recruiting drives but it still can't increase its numbers enough to meet repeated government pledges, a new report finds.
The first thing to say is that recruiting rates in central Ontario are very low by the standards of some other parts of the country. A0A in eastern Newfoundland, for example, had 55 residents join the regular force in 2007, for a rate per 1,000 of 1.15. B1H, in New Waterford, N.S., had a rate of 1.7 per 1,000, and B1P, in Sydney, had a rate of 0.93 per 1,000.
The GTA's highest rate, on the other hand, was in M3K (0.54 per 1,000), which is perhaps a special case, since it contains the former Downsview military base. (Most were reservists transferring to the regular force).
Other than Downsview, the area's highest rate was in Aldershot (L7T) with eight regular force recruits (0.49 per 1,000).
I will admit that the regular force enrolment map did not look at all like what I expected.
Leaving Durham aside for a moment, regular force enrolments in the GTA are concentrated in high-income neighbourhoods more than anywhere else.
In the 416, after Downsview, the highest postal codes are:
M5N, northwest of Avenue Rd. and Eglinton M8W, in southwest Etobicoke M4P, north of Eglinton and Mount Pleasant M4S, south of Eglinton and Mount Pleasant M8X, the Kingsway M6P, High Park and the Junction
In the 905, excluding Durham, the highest postal codes are:
L7T, Aldershot L7J, Acton L7N, south-central Burlington L6K, south Oakville L3Y, Newmarket
By regional standards, regular force recruiting is stronger in Durham, with rates over 0.2 in Whitby, Oshawa and rural areas to the north.
Toronto recruits seem strongly concentrated in the combat arms. M5N's six recruits, for example, were three combat engineers, an infantryman, a supply technician and a logistics officer. A0A's recruits, on the other hand, included 11 truck drivers, four cooks, eight vehicle technicians, six combat engineers and two infantrymen.
 |
| TORONTO STAR FILE |
Leaving aside low-population L0H and L1Y with their one recruit each, reserve recruiting is highest in:
M5S, University of Toronto downtown campus L9N, Holland Landing L0B, Scugog L7T, Aldershot L3Y, Newmarket L7S, central Burlington M3C, Flemingdon Park L7L, Burlington along Appleby Line L4B, Richmond Hill M4E, the Beaches L1T, north Ajax L5H, lakeshore Mississauga west of Mississauga Rd.
 |
| PAWEL DWULIT/TORONTO STAR |
This week's map, showing Ontario's top 20 postal areas for street racing suspension rates, looks very different from the drunk driving map from a few weeks ago.
The drunk driving map was rural, small-town and northern, while the street racing map is strongly centred in the GTA. Only three communities on the street racing top 20 list are outside the GTA and Dufferin County: Petawawa, Aylmer, and Hanmer, north of Sudbury. (The only community to appear on both maps is Petawawa.)
There is a very strong pattern of high street racing rates in an area stretching northwest from Vaughan, covering all of Caledon and Orangeville and stretching into Dufferin County. Only part of this pattern was visible last week.
The street racing age graph is different from the drunk driving graph, with its two peaks in early adulthood and middle age - it simply peaks at 21 and declines.
17% of the drivers suspended for street racing were female, about the same proportion that we saw with impaired driving. Unlike drunk driving, the age graph for female street racers looks more or less identical to the age graph for male ones.
I omitted FSAs with fewer than five suspensions during the period. For the record, here are the areas dropped from the top 20 list:
 |
| AARON LYNETT/TORONTO STAR |
In the Greater Toronto Area, street racing (map) seems to be centred in the area of Peel between Orangeville and Brampton, spilling over into nearby areas of York Region around Kleinburg and Woodbridge.
The one standout exception to the general pattern is M5J, the postal area that takes in the waterfront condos downtown. This is, I hope coincidentally, the neighbourhood across Yonge St. from the Toronto Star. M5J has the third-highest rate in the GTA, and (not to give away next week's map) the fifth-highest rate in the province.
Other than M5J, there is no clear concentration in the 416 area code, other than M2L, in the Bayview and York Mills area.
One interesting thing is that this map doesn't have very much in common with the GTA impaired driving map. Accused street racers seem concentrated in one area, mostly, while impaired drivers are spread all over the GTA with higher numbers in more car-dependent areas.
Low-population L0L, along the 404 in Whitchurch-Stouffville, had only three suspensions during the period but is included on the map out of consistency.
Information was obtained from the Ministry of Transportation under access-to-information legislation.
All the latest news, information and commentary about Toronto on two wheels.
Recent Comments