Problem with track analysis

[fredbull]

Hello, i have a problem. i made some math ,with a friend, to analyse any track i m using.
I use the i2 pro 1.1.6.893
My friend use the i2 pro 1.1.4.456.
Both using the same data and the same math, he gets correct results and i don't.

Here are the maths:

Straight %
stat_mean('G Force Lat' [G] <= 0.5 AND 'G Force Lat' [G] >= -0.5 AND 'Throttle Pos' [%] > 95 , 1, range_change("Outings:Laps"))

Braking %
stat_mean('G Force Long' [G] < -0.3 AND ('G Force Lat' [G]>0.5 OR 'G Force Lat' [G]<-0.5), 1, range_change("Outings:Laps"))

Low Speed corners %
stat_mean(('G Force Lat' [G] > 1 OR 'G Force Lat' [G] < -1) AND 'Speed' [km/h] < 130, 1, range_change("Outings:Laps"))

MEdium speed corners %
stat_mean(('G Force Lat' [G] > 1 OR 'G Force Lat' [G] < -1) AND 'Speed' [km/h] > 130 AND 'Speed' [km/h] < 180, 1, range_change("Outings:Laps"))

High speed corners %
stat_mean( ('G Force Lat' [G] > 1 OR 'G Force Lat' [G] < -1 ) AND 'Speed' [km/h] > 180, 1, range_change("Outings:Laps"))


So as i said, my friend gets good results, and i allways get 0.1% for the medium corners, wich of course is not correct and also makes the other sats false.

I have absolutly no idea what s wrong

[NathanB]

Hi Fred,

It is hard to compare behaviour between 5 year old software and near current software that has 5 years of optimisations and bug fixes applied to it.

There are a lot of things that have occurred between these versions, but in regards to range change stat maths, one item that was resolved in the 1.1.6 branch was the handling of these evaluations.

In versions previous to 1.1.6, if you had a conditional statistic evaluation that took place, with a range change and it was evaluated on one range, but then not evaluated on the next range, older versions would change the value on the next range, even though the conditions for evaluation were not met - this is in fact incorrect behaviour.

I have tested your maths on a piece of data (as yours was not available) and tested against the two versions, with the outcome being the same:

i2 - 114456 array.png (297.26 KiB) Viewed 241 times

i2 - 116904 array.png (286.07 KiB) Viewed 241 times

In 1.1.4.456 and 1.1.6.904 (i re ran this on V893 after this and the outcome is the same) on array evaluation, and when checking a 25 lap range, these report the same.

i2 - 114456 point.png (319.58 KiB) Viewed 241 times

i2 - 116904 point.png (296.07 KiB) Viewed 241 times

Checking this data as a point valuation, again the same condition were met.

I would suggest using a choose equation to sanity check when your maths equation is being evaluated such as in the Forum Maths.xml attached, which contains both your equations as well as the stat mean conditions being evaluated on their on to check when they are being evaluated.

With no screenshots or any information, no further help can really be provided. Workspace variances, such as channel aliasing, and other maths in the workspace could influence outcomes.

Depending on your source data, these equations allow for braking and corners to be on at the same time, and if using real world data, G Force traces will typically be noisy enough that these conditions will come in and out of evaluation multiple times through one of these regions.