Page 1 of 1
| Failure probability does not work in intuitive way 10/03/2015 at 21:34 #69937 | |
arabianights
138 posts |
Try this: set track circuit failures to a low probability of occurring, and point failures to a medium chance of occurring. I bet you'll get more track circuit failures. I *suspect* the cause is one or both of two things - either TCFs are more likely to happen because they are more likely to occur in real life than point failures, or (and I expect this is it) a probability of medium is, say, a 1 in 500 chance that a failure will occur in 30 minutes to any paticular track circuit/signal/point/whatever. But of course the problem is that there are far more track circuits than there are points. To me, a much more intuitive way for those sliders to work is that equal levels mean there is equal chance that something will occur in the game (and thus for an individual track circuit, less probabilty of a failure than for an individual set of points) . I suspect this would be the same for most people. Last edited: 10/03/2015 at 21:36 by arabianights Log in to reply |
| Failure probability does not work in intuitive way 10/03/2015 at 21:52 #69938 | |
jc92
3690 posts |
Failures are based on a percentage chance per switch of point/occupation of TC/change of aspect rather than per hour I think
"We don't stop camborne wednesdays" Log in to reply |
| Failure probability does not work in intuitive way 10/03/2015 at 22:28 #69939 | |
|
clive
2789 posts |
I explained exactly how failures worked a few weeks ago. I'll leave it to someone else to find the thread.
Log in to reply |
| Failure probability does not work in intuitive way 10/03/2015 at 22:55 #69940 | |
|
jc92
3690 posts |
http://www.SimSig.co.uk/index.php?option=com_kunena&view=topic&catid=14&id=39138&Itemid=0#68923
"We don't stop camborne wednesdays" Log in to reply The following user said thank you: arabianights |
| Failure probability does not work in intuitive way 11/03/2015 at 00:01 #69943 | |
|
arabianights
138 posts |
" said:I explained exactly how failures worked a few weeks ago. I'll leave it to someone else to find the thread.I am not asking how they work, at least not primarily. I am describing how I believe they ought to work from a ux perspective. The method you describe could easily be weighted so that the same positions in the scroll bars correspond to roughly the same number of failures occuring in a typical session. If that is not a goal then I question the point of scrollbars - maybe numeric value textboxes would be better. All IMO of course. Last edited: 11/03/2015 at 00:05 by arabianights Log in to reply The following user said thank you: DonRiver |
| Failure probability does not work in intuitive way 11/03/2015 at 01:17 #69944 | |
GeoffM
6377 posts |
" said:The method you describe could easily be weighted so that the same positions in the scroll bars correspond to roughly the same number of failures occuring in a typical session.That assumes that all sessions are typical. Hint: they're not. Just one example: years ago I was involved in a study to formulate timetables in such ways as to reduce point movements - keep everything following each other as much as possible. To a certain extent that has been achieved with both the WCML and GWML: as little as possible switches lines. If you go back to even just the 90s there were a lot more trains switching lines. Point failures thus vary for the same simulation with different timetables. " said: If that is not a goal then I question the point of scrollbars - maybe numeric value textboxes would be better.No! Once you give numbers then that gives users a number to aim for and then they start complaining that "I specified X but actually got X+1". Scrollbars give the vagueness that's required from HCI. SimSig Boss Log in to reply |
| Failure probability does not work in intuitive way 11/03/2015 at 18:38 #69953 | |
|
arabianights
138 posts |
You don't have to do it like that. The vast majority of movements are where train moves from signal a to signal b via a route set. All you need to do is have a probabilty of a failure of point/signal/track circuit in one movement, then pick one of the points/signals/track circuits affected by the movement. You can then use the current model for other movements e.g. PSADs. Log in to reply |
| Failure probability does not work in intuitive way 12/03/2015 at 18:43 #69984 | |
|
arabianights
138 posts |
I realised the problem in the algorithm I described above when driving home from work tonight, no need to point it out :lol:
Log in to reply |