absolute value threshold for variances

[nicolebottoms]

if there is data in the system with a variance, i want the system to ignore, or at least flag differently, or not at all, variances with an absolute value threshold of under 20 or some appropriate number not sure exactly what it should be.
for example when i'm looking at a variance report i'm checking the high numbers. chilliwack lake has a 700% variance for sscfe campers in may.

but in reality, there are 8 campers, up from the average of 1 camper. piddly numbers like this aren't a big deal, i want the big variances of an extra 300 people.

[LindsayMacfarlane]

January 22, 2025: Identified has a high priority.

Mural board notes and discussion.

[LindsayMacfarlane]

May 14th A&R check-in with PROS team. This item continues to be flagged as a top priority for their team to resolve.

[Dianadec]

Todo: Refine a bit more - @LindsayMacfarlane to discuss with @davidclaveau

[davidclaveau]

Some general thoughts:

• Variance reports currently flag large percentage variances even when the numbers are very small (e.g., from 1 camper to 7 campers results in 700% increase)
• The we're looking for more significant changes, so big variances in volume
• Small changes in low-volume items or locations should either be ignored or flagged differently

Implementation thoughts:

• We should decide on a threshold for ignoring / flagging small variances.
  • e.g. abs(actual - expected) < 20 treat as “low significance"?
  • Threshold could be configurable, maybe stored somewhere by devs/users.
• Still calculate percentage, but flag or filter those below the absolute threshold.
• Do we want these low-significance variances completely removed from the report?
• Should the threshold be global for every metric?
• Is the threshold always an absolute number or should it sometimes be relative (e.g. ignore if < X% and < Y absolute)?

[cameronpettit]

I don't think this ticket is ready to action. We need to be clearer on what exactly we would like to see implemented. I like all of Daves suggestions above, but we need to decide which to implement and how.

Having a single, unchanging threshold for significance and excluding data from reports that don't reach the threshold is one thing.

Having variable, configurable thresholds that depend on the field, activity, park, administrator, etc... much more complicated to implement.

Depending on which path we take, I wonder if it wouldn't be easier to do this kind of filtering in Excel after the fact?

[sallyhaynes]

@cameronpettit @LindsayMacfarlane @nicolebottoms The threshold would not need to depend on the park or administrator, probably just the field. I don't think we want to defer this to excel- the first issue is while Nicole and I don't have trouble formatting in excel, the absolute values are not currently present in the variance report to filter (this has been requested in a separate ticket). The main issue though is that we want to remove as much friction from using the variance warnings and variance report as possible and we have received complaints from the Park Operators that the variance reports contain too many inconsequential variances and therefore take longer than necessary to sort through. Currently when I send variances to the PO I have to have the A&R export along side it to see which variances are due to small numbers and delete them out before I send it to the PO. We want the PO to be performing these checks themselves (and some have little to no excel skills), so making it as easy as possible is important.

[cameronpettit]

@LindsayMacfarlane @sallyhaynes @nicolebottoms okay, so this has been identified as a must. If the significance of the variance depends on the field, we would need a list of rules - probably minimum one per field - that describes the situation in which we would not consider the variance at all.

Activity	Field	Variance trigger %	Variance trigger absolute value
Frontcountry Camping	campingPartyNightsAttendanceStandard	>20%	>100 nights
	campingPartyNightsAttendanceSocial	>20%	>30 nights
	campingPartyNightsAttendanceSenior	>20%	>30 nights
	campingPartyNightsAttendanceLongStay	>20%	>30 nights
...	...	...	...
Day Use	peopleAndVehiclesVehicle	>20%	>500 vehicles
...	...	...	...
Frontcountry Cabins	...	...	...
Backcountry Cabins	...	...	...
Backcountry Camping	...	...	...
Group Camping	...	...	...
Boating	...	...	...

Using the above example, the variance would only be shown on the map if it satisfies BOTH the variance trigger % AND the variance trigger absolute value. Is that correct?

Would the variance absolute value be calculated by comparing to the previous year or the average of the past three years?

If so, we will probably have to create a ticket to capture the creation of a fully fleshed out table of values.

[nicolebottoms]

@cameronpettit yes, needs to satisfy both conditions.
the absolute value would be calculated by comparing the past 3-years average, like the regular variance check.

here is a quick one of my thoughts, please weigh in @sallyhaynes

<style> </style>

Activity	Field	Variance trigger %	Variance trigger absolute value
Frontcountry Camping	campingPartyNightsAttendanceStandard	>20%	>100 nights
Frontcountry Camping	campingPartyNightsAttendanceSocial	>20%	>30 nights
Frontcountry Camping	campingPartyNightsAttendanceSenior	>20%	>30 nights
Frontcountry Camping	campingPartyNightsAttendanceLongStay	>20%	>30 nights
Frontcountry Camping Winter	campingPartyNightsWinterAttendanceStandard	>20%	>50 nights
Frontcountry Camping Winter	campingPartyNightsWinterAttendanceSocial	>20%	>30 nights
Day Use	peopleAndVehiclesVehicle	>20%	>100 vehicles
Day Use	peopleAndTrail	>20%	> 100 people
Frontcountry Cabins	campingPartyNightscabins	>20%	> 30 nights
Backcountry Cabins	adultsbackcountrycabin	>20%	> 30 adults
Backcountry Cabins	youthsbackcountrycabin	>20%	> 30 youths
Backcountry Cabins	familybackcountrycabin	>20%	> 30 families
Backcountry Camping	peoplebackcountry	>20%	> 50 people
Group Camping	standardgroupnights	>20%	> 30 nights
Group Camping	standardgroupadultpeople	>20%	> 50 people
Group Camping	standardgroupyouthspeople	>20%	> 50 people
Group Camping	standardgroupchildpeople	>20%	> 50 people
Group Camping	youthgroupnights	>20%	> 30 nights
Group Camping	youthgrouppeople	>20%	> 50 people
Boating	boatingnightsondockboats	>20%	> 50 boats
Boating	boatingnightsonbuoysboats	>20%	> 50 boats
Boating	boatingmiscellaneousboatsboats	>20%	> 50 boats

[sallyhaynes]

@nicolebottoms Thanks for putting this together. I've got a couple suggestions but want to test out some of my thoughts with the math to make sure it makes sense before putting them forward. :)