h a l f b a k e r yLike a magnifying lens, only with rocks.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
traffic forecast
service uses special event info, and traffic records, not just real time info | |
We have seven day weather forecasts, why not traffic forecasts too? Accidents are hardly the only thing causing traffic. There are really obvious things that can be known sometimes years in advance: special events, planned closures and weather. There is also historical traffic information to take
into account (eg. There is a particularly heavy day before and after Christmas because all government jobs have the same vacation begin and end days.) If there is a freeway that is used by suburban dwellers to get to a popular lake, and a warm day is coming up, the traffic forecast will take this into account. Also, most freeways have different peak times for their opposite directions. Most freeways are not jammed at the same time in both directions, but if you are new to the route, you can't be sure which way is heavy in the morning/afternoon. These are all things you can find out about far, far in advance if you went to about 100 different websites, and spent hours taking all the info into consideration.
Here's how it would be automated. A script scours relevant web pages for events and times they occur and parses the info, reading the numerical data for locations and times of the event into variables which are then used by functions which differ from freeway to freeway. The rest becomes trivial.
An epistemic community consisting of myself, and some trusted individuals around the US and elsewhere could oversee the service, and add those big things that might not be listed anywhere predictable (eg. president arriving at the airport, and shutting down the freeway that gets him to the convention center).
Volunteers could also amend the info, but only after thousands of people say the same thing will it become true, and be taken into account by the program.
You could also just hire cheap labor in another country to do it. Infact, this could be a great way to get around captcha if you are a spammer...oops, someone mark this for deletion.
[link]
|
|
//Way baked for Smartphones.// |
|
|
nah. Those are all hyper realtime programs. |
|
|
Is there even a place where you could see what traffic was like on this day last year? I sure as hell can't find one.
So my program will also store all of this data so that it can use it as part of the overall traffic calculation. |
|
|
If you could bake this -- even a version which could predict traffic a few hours into the future -- it would be a wonderful addition for road navigation systems. |
|
|
Some current systems use real time traffic data to plan routes that avoid roads that are delayed... but they only know what delays are "at present." |
|
|
If it's an hour before rush hour begins, and you want to plan a two hour trip, that real time traffic data is going to be invalid halfway through your trip. |
|
|
If you had a predictive model to aid the system, it could plan a trip so that when rush hour starts, you'll be on a side street which never gets hit by traffic. That side street might be one which is slower than the highway when the highways are clear, but faster than the highway when they're covered with cars. |
|
|
//Is there even a place where you could see what traffic was like on this day last year?// Why? Are the accidents or breakdowns likely to be in the same places? [-] |
|
|
//Are the accidents or breakdowns likely to be in the same places? // |
|
|
I think this is a great idea. Even on a fairly simple level, it
could help. For example, traffic patterns where I live
depend very much on whether it's school term and
university term. Throw in data on major events, public
holidays and so-on, and I think you could make quite a
good prediction of traffic densities. Accidents are of
course unpredictable, but they're just noise superimposed
on the trend. |
|
|
If someone like Garmin hasn't considered this, I'd be
surprised. |
|
| |