
Slashdotted... I mean, El Reged...
It doesn't roll off the tongue quite so well, but the site is definitely struggling.
Web coders have plotted a map of UK road crashes using data pulled via the Department of Transport's API. Launched on 27 September, CrashMap is the first application to make use of the Road Accidents and Safety dataset that was released on 30 June this year. CrashMap is listed on the data.gov.uk site as "the only up-to-date …
Well, the site seems to be down, so I can't comment from experience yet...
Still, I suspect it would prove an "enlightening" exercise were our government IT chaps and chapesses to compare and contrast both the results and the costs of both this and the crime data site.
If they look carefully there is probably a couple of lessons involving "overpaid consultants" and "letting people who know what they are doing manage the project themselves" to be had in there somewhere...
If this was a fun project then good luck to them.
But if this is some sort of business demo then why is it not aggregating data so I know where to zoom in to? It sits there for ages then says "5000 results, zoom in". Google provides the functionality in the api. And why can't I click to zoom? Reposition, click the zoom button a few times, repeat 5 times, arghh.
Impressed that they've managed to get rid of Google's copyright notice though. Not sure what Google will think about that.
There is a junction not far from where i live that has an accident a day relatively speaking (minor and major).
So notorious is it that the local authority are sending out letters saying "due to the statistically high number of vehicle colliosions and / or injury on xxxxxxxxxxxx road" they are totally redesigning the junction. However, not a single incident is showing up on that site.
So, where have these fucking numpties dredged this pile of irrelevant innacurate data from...
Bollocks offline and just as much bollocks online....
Every time you move the map, even slightly, it clears all the data, re-queries the database and redraws the pins. I'm no programmer, but surely there must be a way of caching the data you've already got? Then it'd only have to go back to the database for the sectors that weren't on the map before, and, while it's doing that it could at least redraw the pins it already knows about, which would make it look less terminally slow and ought to reduce the load on the server.
STATS19 the 'crash' database has been known for a long while to be a pretty poor (=completely) record of cycling incidents. Hello other road users. Now you know for yourselves. Perhaps harassing a bunch of government wonks/politicians is in order ... ...Oh you can only be bothered to rant on el-reg not get at the public servants - sorry I wasted your time.
Considering that most accidents are not formally reported, this effort is a complete waste of time.
Bad drivers remain behind the wheel to go on and cause their next accident.
Reporting accidents should be compulsory. Unfortunately, they won't be. Its not like the police actually cared or anything, just as long as they do what they're told to do....
They have missed at least two fatal road traffic incidents on the Aldersbrook road, E11 and E12 but they can only be as accurate as the data set they are using.
Crashes are a major cause of congestion with all the associated fiscal costs. Understanding where incidents can only help us in preventing future incidents.
To look at detailed information, you must register. A process which supposedly requires only the information "necessary" to look at the information.
Further examination shows that this so-called "necessary" information requires an email address, a first name and a surname.
All the information comes from a public domain database and the information is presumably read-only so there's no ability to make changes, and therefore no reason to need to know who is accessing the information. Consequently, the only information that can be truthfully "necessary" to look at the information is NONE whatsoever.
Precisely what the operators intend to DO with the information they are collecting I cannot guess, but there's certainly no good reason to need to know anything other than to count the number of incidents examined, which can be done in the server without requiring any personal information.
"Precisely what the operators intend to DO with the information they are collecting I cannot guess, but there's certainly no good reason to need to know anything other than to count the number of incidents examined, which can be done in the server without requiring any personal information."
If you bite and register, the next screen promptly informs you of the number of 'credits' on your account (zero) and you are given the opportunity to purchase credits to allow you to access the accident report.
I, equally promptly, set the first and surname on the account to 'Fuck' and 'Off', and left the site - not because I have any objection to them charging for value-added aspects of their service, but for the shitty, greasy, gutter-trash trick of not giving the charge and cost details upfront but rather extracting personal details first and then yelling "Ha-Ha-HA! Gotcha!" like some street-corner shysters.
The government data only shows the RTIs they know about. In other words it doesn't include an awaful lot of damage only accidents.
Our local authority has a serious problem with this. They have told me that the only figures they can get from police and central government are generally for KSI incidents. The idea of a road safety program is surely to prevent KSI incidents, yes? Well the boundary between a KSI incident and a damage only incident can be tiny, particularly in a low speed accident. All it takes is a vehicle occupant to be over or undersized*, a poorly maintained vehicle or any one of a number of other apparently insignificant changes.
A particular junction could be constantly strewn with the glass of minor incidents and the local authority might not be aware that this is an accident black spot. Then by freak chance a handful of people get injured and suddenly the LA are forced to do something about it. The problem with the stats is that they are only any good for closing the stable door after the horse has been hit by a truck.
* It's true, most vehicle safety devices are designed to cope with people of around average height and weight. If you are tall or short or fat or thin then you stand a greater chance of being injured in an RTI.
I'd like a map of the traffic volume divided by the accident rates. Ideally one with the roads coloured by their accident/flow ratio.
There might be a lot of accidents on the M25, and very few in the far north; but there are millions thundering around the M25 every day vs perhaps only a few hundred trundling around the far north.
Some years ago I set out to do the same thing using the dats available for 1991 to 2003 (iirc). The quality of the records available over the period was not great and the data structures and what was bieng recorded changed a little over the years. I tweaked most of the evident anomalies or elected to ignore or summarize the most problematic fields.
I could get the map references to add pins to the map system easily enough but noticed that some record seemed to be in the wrong place, or missing entirely.
Digging deeper I spotted cluster of 'accidents' in the middle of nowhere, particularly in the North of Scotland. Then I discovered the 'Off Map' accidents in the North Sea and quite a few in the Atlantic and the English Channel. Clearly these anomalies could be repeated 'on land' as well. And they were. Many records were out of position - by much more than 10 metres
With several million records to assess and no honest way to correct the anomalies without access to the original raw (paper based) reports (and even then no guarantee that the information noted was correct) I concluded that to go further was a waste of time. One can live with the odd anomaly but as I recall the numbers of obvious anomalies for location alone were quite significant and adding in other suspicious information made the whole exercise questionable.
The more recent data may be more accurate than the earlier stuff. For example I assume that more, maybe all, of the reports are captured electroincally rather than written and transposed and that taking a map reference form the GPS device requires less 'skill' than reading a map to get the reference. (On the other hand in the very north of Scotland and the Shetland Islands -again from memory - the aggragated database struggled because the relevant fields could not store the required latitude values. Not encouraging for the rest of the design quite frankly.)
All in all with no realistic chance of applying 'corrections' I thought it unwise to take the project further. It sounds like more recent data is no better. I'll have a look at some point. Maybe.