2/24/2010 5:08:04 PM
MJ
Posts: 14
|
Hello
Can anyone explain to me about the 'Calculation Paramters' under settings?
What exactly does the 'PT Sampling Interval (min)' mean? I have changed this from 10 minutes to 5 minutes and even 1 minute in some instances and achieved different results from my network accessibility isochrone output, what does changing this setting actually do?
Also, with the following parameter 'Maximum walk time to PT - local (min)' the name suggests to me this only applies to local accessibility calculations, am I correct? Or does it limit the amount of time an individual will walk along the road network within a network accessibility calculation? As if it does then this could get round the previous problem I have been involved in discussing regarding whether the road network should be included in an Accession calculation
MJ
edited by MJ on 2/24/2010
|
|
permalink
|
3/26/2010 2:19:22 PM
RobT
Posts: 36
|
Hi MJ,
My understanding of PT Sampling Interval is how often a new "journey" is undertaken when using PT. So across a 2 hour period 12 journeys will be done with a 10 min interval and 24 journeys with a 5 min interval. I believe that a 10 min interval will do the following . . .
Leave Origin @ 10am
Walk to PT stop (2 min walk arrive at 10:02)
Bus arrives at 10:07 (5 min wait)
Go through PT/Road network to destination
Arrive at destination at 10:25
This would give a 25 minute total journey, now if a 5 minute PT Sampling interval was selected then the quickest journey would be to leave at 10:05 and arrive at the PT stop as the bus arrives - thus eliminating 5 minutes from the journey. Can anyone clarify this.
Cheers,
Rob
|
|
permalink
|
4/8/2010 4:39:40 PM
keithdrew76
Posts: 44
|
Rob
My understanding is that in principle you are right, but actuall Accession works backwards from destination rather than origin. Thus wait time is equal to sample time (at dest) - time getting off stop - walk time from destination to stop. So if end your journey at 10, walk for 2 mins and get off bus at 955, wait time = 3 minutes. This was done back in 2004 to keep calculation times down (algorithm)... You effectively have no wait time at origin stop...
Hope that messes your head up like it did to me :-)
|
|
permalink
|
4/12/2010 12:31:21 PM
RobT
Posts: 36
|
Thanks Keith, so would it be best to use a 1 minute PT Sampling interval? This will then minimise the last wait time?
Rob
|
|
permalink
|
4/13/2010 9:32:21 AM
keithdrew76
Posts: 44
|
Hi Rob,
Personally I would consider ticking the 'ignore first wait time' (or what ever the text is) box to ignore this 'end' wait time, rather than 1 minute sample (keeping to a 5 minute sample). Thus you are left (or should be left) with wait time from any interchange only. You could do some kind of analysis at your starting point (origin) to see how much wait time is applicable to that area, or you could simply add a notional time (although this will impact on your travel times!). This method is what I propose in the presentation in the 'user work' folder.
If you set yoursampling to 1 minute your calculation time would be increased significantly (10 fold from a 10 minute sample) - may not be an issue if your calcs are not too big though...
Hope that helps!
Keith
ps if anyone thinks I wrong with the wait time calc method please shout at me :-)
|
|
permalink
|
4/13/2010 3:45:22 PM
RobT
Posts: 36
|
Am I correct in my thinking that to do the ignore first wait time you have to not use the road network?
Rob
|
|
permalink
|
4/13/2010 3:58:38 PM
keithdrew76
Posts: 44
|
You are right - its not something I've used to be honest (I either calculate my own wait time inc interchanges in GIS or use the measured values by Accession at 5 minute sampling)...
Help file gives the following statement...with a road network it suggests a sample time of 1 minute (as you suggested) is equivalent. Hopefully your calc isnt too big with this setting :-)
*********************************************************************************************************************************************************************
The No First Wait Time option is available when doing a PT Only, Outbound, Time based calcualtion only.
The algorithm ignores the first wait time encountered (from the Destination end) if the wait time is less than the Sampling Interval. This feature allows for an optimum setting of a 5 minute sampling interval to create results that have an accuracy closer to a one minute sampling interval. Thus keeping the calculation times low but enhancing the accuracy of the journey times. As with all efficiency techniques there is a trade off and in some cases ACCESSION will not calculate the fastest journey between two points. To fully understand when this happens it is recommended that you perform a calculation for your network with a one minute sampling interval and compare this to a five minute no wait time calcualtion.
|
|
permalink
|
4/13/2010 3:59:16 PM
keithdrew76
Posts: 44
|
Sorry about the blanks, highlight the text to unravel :-)
Keith
|
|
permalink
|