Interpolation of data

[sdezego]

Just a quick question as I am using 2.61 for tuning via datalogs. How do you interpret data points and more importantly handle any outlying or odd data points.

I have not figured out why, by my logs have some bad datapoints sprinkled here and there (nothing major) and I don't want to cause big offsets in calculations. I see that the Hit rate and such plays a factor so, statistically these point are probably diminished, but I just want to make sure.

RPM channel is perfect, but for instance my Spark log as a bad data point every once in a while and a value that is not part of an overrun and not part of any valid map point. I had some issues with A/F via the WB02, but found that to be a poor connection in the 2.5mm plug coming from the LM1's analog output

I know that ML is just reading the datalog and is not responsible for the data itself, but I was just curious about thresholds for data averaging.

Shawn

[jsmcortina]

What firmware version are you running?

James

[LT401Vette]

I'm guessing you are talking about VE Analyze
There is some detail on how it works and filters data here:
http://www.ideasandsolutions.biz/MegaLo ... lysis.html

Erratic spark in a log would not cause a problem. Bad data from RPM, Gego, AFR or O2 would certainly hurt results.

Any records where Accel/Deccel, Over-run, warmup or not running will be filtered and not looked at during analysis. You can also set any of the filters in the Advanced box to set your own filters if say you don't want idle or WOT looked at.

[sdezego]

Yes, VE Analyze. This is exactly what I was looking for.

As for the the spark spikes, I realize that data is not used, but it spawned my thoughts as to other areas while looking at my data. Turned out it was 029T. corrected with firmware 029v.

I'll do some more reading.

Thanks!