Page 1 of 4

Historic Data

Posted: Thu Oct 11, 2012 10:55 pm
by
When I tell Waze to take me from my home to my work in the morning, it appears to not take historic data into account.

For example, let us say I have 2 routes I could take to get to work. Route A and Route B.

Route A is the route I would take when I know there is no traffic (night, weekend, holiday, etc...). When there is no traffic, or when traffic is flowing at optimal levels, this route takes me 35 minutes. On a typical work-day this route takes 55-60 minutes.

Route B is the route I take on a normal day. Route B when there is no traffic takes 45 minutes. On a typical work day this route takes 50 minutes.

Route A, during rushhour is ALWAYS slow. Additionally, there is a huge on-going construction effort on this route that always causes additional delays. In 2 years of taking Route A during a normal work day, I have NEVER had it take less than 55 minutes to complete the trip.

For a week now, at 7:15am, I tell Waze to take me to work. Every morning Waze wants to take me down Route A. Every morning it tells me Route A is going to take 35 minutes. This is wrong. During the work week, Route A will NEVER be this fast. Route A is NEVER the fastest route.

Does Waze not take historic data into account at all? Waze should know that by 7:30 most of Route A will be heavy traffic and the commute time will increase accordingly. Why should I keep using Waze if it insists on guiding me down the slowest route. If I have to constantly ignore what waze is telling me....whats the point?

Re: Historic Data

Posted: Fri Oct 12, 2012 2:37 am
by AlanOfTheBerg
How often have you driven route A, purposely getting caught in the slower traffic so that Waze learns that this route is slower? Just because it might know how fast Route B is doesn't mean it has the data about route A in order to make the right choice.

Re: Historic Data

Posted: Mon Oct 15, 2012 2:49 pm
by AlanOfTheBerg
sublimese wrote:Why do I need to drive route A? Isn't the entire point of Waze that we are crowd-sourcing traffic? If route A is currently full of waze users, does waze really need me to also drive down route A so that it can tell me that that route A is bad?
Not necessarily, no. If you can see that Route A is "full" of Wazers, I would think it would have valid data. I didn't see that you mentioned this about the other route earlier, hence my suggestion, but now that I think about it, with A being I-25 in Denver, I'd think the data is fine.

I know there are several editors in the Denver area, so that road should be just fine, as should route 'B' in general. Route length is something which can affect the road type which Waze will prefer to use which we have no influence over. At certain distances (I cannot recall specifics at this time), Waze will try to not route on anything less than primary street except to get you to one, and off of one near the destination. Longer distances and even Primary Street isn't preferred over highway/freeway. Your list of streets looks like highway or greater for the middle sections, which, again, should be fine for routing even for longer routes.

What is the total distance for your route B? Chatting with bgodette about it may be of use too, as he should be pretty familiar with the area.

Re: Historic Data

Posted: Mon Oct 15, 2012 5:28 pm
by AlanOfTheBerg
chrysrobyn wrote:(I wish Waze had a helpful way of permalinking a route, including waypoints.)
You can in Livemap. Well, you can permalink the origin/destination, but not the route itself. That may be future functionality in the app, though, having Waze actually remember your preferred route.

http://www.waze.com/livemap/?q=denver,% ... &to_seg=-1

Looks like this:
http://imageshack.us/a/img835/7660/denverroute700px.png

Re: Historic Data

Posted: Mon Oct 15, 2012 5:38 pm
by AlanOfTheBerg
Has it been asked yet (and answered) whether your Route B shows as an alternate route in the app? I've seen odd behavior sometimes where, for whatever reason, the route Waze chooses isn't actually the fastest. (You do have "fastest" selected in the app settings, right?)

Re: Historic Data

Posted: Mon Oct 15, 2012 5:48 pm
by AlanOfTheBerg
sublimese wrote:When I tell it to show me the optional routes, it shows 3 additional routes, all of which are variations on Route A. All of which show best-case commute times, not current actuals or historical times.
That's disappointing, but if Waze, for whatever reason, thinks those are the fastest, that makes mathematical sense. But the question remains, "Why doesn't Waze take into account the historical/known data which shows that route is actually slower than others?" such as your Route B.

You've got a lot of good info in this thread. It may be worth opening a ticket to see if you can get a direct response. Use the Quick Help box on the Support page, enter a "subject line" then click the Email link. Fill in the info, include some text and a link to this thread.

Re: Historic Data

Posted: Thu Oct 11, 2012 11:00 pm
by AndyPoms
Have you checked Route B for map errors?

Re: Historic Data

Posted: Fri Oct 12, 2012 6:02 am
by bgodette
I'd like to know what route B is because every single destination I've set inside the 470/nwp loop and into central Boulder has been accurate to +/- 2 minutes even during rush. The only times I've had it be inaccurate is when an accident has just happened or there's something that doesn't happen at the same time on the same day of week (sports, presidential debates, stolen tanks, etc).

It's all my phone's fault for using Tapatalk.

Re: Historic Data

Posted: Mon Oct 15, 2012 4:00 pm
by bgodette
AlanOfTheBerg wrote:
sublimese wrote:Why do I need to drive route A? Isn't the entire point of Waze that we are crowd-sourcing traffic? If route A is currently full of waze users, does waze really need me to also drive down route A so that it can tell me that that route A is bad?
Not necessarily, no. If you can see that Route A is "full" of Wazers, I would think it would have valid data. I didn't see that you mentioned this about the other route earlier, hence my suggestion, but now that I think about it, with A being I-25 in Denver, I'd think the data is fine.

I know there are several editors in the Denver area, so that road should be just fine, as should route 'B' in general. Route length is something which can affect the road type which Waze will prefer to use which we have no influence over. At certain distances (I cannot recall specifics at this time), Waze will try to not route on anything less than primary street except to get you to one, and off of one near the destination. Longer distances and even Primary Street isn't preferred over highway/freeway. Your list of streets looks like highway or greater for the middle sections, which, again, should be fine for routing even for longer routes.

What is the total distance for your route B? Chatting with bgodette about it may be of use too, as he should be pretty familiar with the area.
The data on 70, 470, 25, 285 (Hampden) and 85 (Santa Fe) will be fine, and like I said Waze is consistent in +/- 2 minutes for my daily which is SE Aurora to Downtown. All site visits have been +/- 2 minutes to where my "driven" area covers pretty much all of the metro. Based on the description of Route B it might be running into a traveled vs straight line as all parts of B other than Kipling north of I-70 and Happy Canyon (Primary) and the actual destination are Minor or higher.

There is also the unknown about how far back the historical averages go, and how much weight any newer data has if any.

Re: Historic Data

Posted: Wed Oct 17, 2012 4:48 pm
by bgodette
chrysrobyn wrote:My money is on the routing server not agreeing that 285 / Alameda Ave is a valid cross-town route.
No, not even close.
chrysrobyn wrote:I think you'd have to upgrade it to a Major Highway in order to overcome that barrier
That would do nothing. Type only comes into play if there's a lack of historical data, which there isn't, or the route is very long (>200 miles before Major vs Minor starts to matter), which it isn't, and even then the distance tiers only increase the likelihood of the route getting pruned, it doesn't outright exclude Types.

The probable issue is one of actual distance vs straight line.

Re: Historic Data

Posted: Wed Oct 17, 2012 6:31 pm
by bgodette
chrysrobyn wrote:If that's the case, is there a next course of action for the user or for his AMs?
Yes and no. We've managed to get Waze to increase the ratio at which is drops a route. In at least one distance tier it seems to be about 6:1.

My own commute in Denver is roughly this, barring accidents that happening along the route after starting, it has been accurate to +/- 2 minutes regardless of which side it decides to send me.