Record of filter performance

[Falcon]

Hi Stan,

The Custom Analyzer is a great tool for displaying a filters’ success or otherwise over past draws. My approach and I presume used by others is that if I have a complex filter that failed at the target draw I then look for the faulty element within the filter and make a correction to it. When I then run the Custom Analyzer again I effectively get a clean sheet showing all is normal. However what is then lost is what has been the overall performance of each filter over the many times they are used. Is it failing every 2nd draw or every 10th draw or every 50th draw? If one wants this sort of information one needs to keep an external bookkeeping record. Thus might a complex filter be able to store its past performance? Could there be a tick box in the complex filter which would record that the filter failed at the current draw? And then when the past filter performance is displayed in the Custom Analyzer table at each draw where the filter has previously failed the cell might be displayed in a different colour?

thanks

falcon

[dreamweaver]

Hi Stan,

I came across this post and I too have similar processes in evaluating past performances of various Complex Filters and being able
to look at Historical performances of those Complex Filters at one's disposal. I feel what Falcon suggests has some usefulness and
value for interpreting results with some ease.

I think if this feature would be available it would be of value.

Cheers

[stan]

Technically this would be a problem as the past results must be stored somewhere. The actual filter can be applied in more than one lottery so it can't be in the filter definition file and there's no unique filter identification so it's hard to match some stored results to a complex filter file.
Besides if you 'fix' the filter after each failure then any stored past results will apply to a different filter setup, won't they?

[Falcon]

Hi Stan,

Well I suppose one might say it seemed like a pretty good idea at the time.

I do concede that there certainly are some issues. Perhaps I could hope that a second record set within the program might allow other features to be eventually incorporated and that the mapping of filters to their respective lotteries and also their performance record be an initial building block.

Whilst filters do fail and they become ‘fixed’ this is actually the issue I believe should be tracked – the frequency of failure. A high frequency obviously means the filter design – selection of parameters etc is deficient and so the structure ought to be re thought through. And perhaps even the converse for a never failing filter – might it be adjusted to work a bit harder and so actually have some sort of acceptable overall fail rate?

Thanks

falcon

[stan]

Alright, but such booking keeping would have to be manual process - the filter can be applied/previewed many times over and the application cannot know when it's time to record the failure and when you're just testing new settings.
And I still need to come up with some reasonable user interface to manage such records.