Matrix

by hn

Downloads: 2,336 (3) • Reviews: 0

Version:

0.7.6.9814

Updated:

Jul 29, 2011

Matrix

Postby gerhard » Mon Aug 23, 2010 1:50 pm

GaryS wrote:I would opt for something even more simple, keeping the direction the same. A1-B1, B1-C1, A1-C1 would all be 'outbound' pairings. I can do another route for the return.

The clear issue to avoid is having Matrix confuse an A1-B1 pairing with an A1-B2 pairing. That's when I see times and distances skewed in the sample of all activities.

There seem to be an issue with how to detect that one passing is unique. It so much easier for the human eye to see such pattern compared to a program. So what you determine is A1, the program sees as A1', A1'' etc.
I am sure the detecting can be done better, but will need information about the problem, unless I find an own activity with the problem.
Please open an issue on CodePlex with an activity and Preferences.System.xml or PositionGroup file with the name of the Trail.

GaryS wrote:However a better suggestion might be to use Trails to define the circle AND define the expected order of points for use in Matrix. For example, if Trails knows that A-B is 1 mile, then apply some margin of error, say 10% and any A-B pairing over 1.1 mile cannot be A1-B1 but instead must be A1-B2. If the plugin can do this automatically, that would be optimal. Manual tweaking would just need to be done if no A points (or B, etc) fall in the circle which will be obvious since the Trails entry will be empty.


To require an order is possible, but that would be against the Matrix principle that Matrix was designed to do.
To implement checking of "legs" would be a complicated task, unless the bird distance is used, is not feasible.

In this scenario, a multi activity Trails would be a better and simpler solution.
gerhard
Contributor!
Contributor!
 
Posts: 1798
Joined: Mon Nov 27, 2006 7:07 pm
Location: Sweden

Postby gerhard » Mon Aug 23, 2010 5:10 pm

A test build with increased hysteresis for "starting new search" for a point from 1.2 to 4 * radius.
http://hem.bredband.net/gerhardnospam/t ... .st3plugin
Some other stuff changed too (comboboxes replaced, some internal changes).
gerhard
Contributor!
Contributor!
 
Posts: 1798
Joined: Mon Nov 27, 2006 7:07 pm
Location: Sweden

Postby GaryS » Mon Aug 23, 2010 5:30 pm

gerhard wrote:A test build with increased hysteresis for "starting new search" for a point from 1.2 to 4 * radius.
http://hem.bredband.net/gerhardnospam/t ... .st3plugin
Some other stuff changed too (comboboxes replaced, some internal changes).


Looks like the multiple A1-B1 entries are gone, as are the A1-A1 entries. Great!
GaryS
Donated!
Donated!
 
Posts: 77
Joined: Thu Jan 21, 2010 6:25 pm

Previous

Return to Matrix

Who is online

Users browsing this forum: No registered users and 1 guest