Actix - which binning mode suits best?

[serbesi]

hi there,

i've been confronting various types of binning modes in actix lately, and i've seen the results differ quite a bit from mode to mode. Time binning (1000ms) was my default setting, but when i switched to Message binning (1), the results differed 7-8% in each category. Then i tried Time binning (500ms) and the results were similar to message binning. Logic suggests that the smaller the binning time, the more accurate the results, at the expense of processor and memory load. What do you think could be a good trade-off between accuracy and processor load?

Thanks very much

[gigel]

depends on what you are looking for:

if you want just to post-process and analyses a drive test use geographic binning you can use 50m in-city and same or higher out-city
if you want to analyses the signaling use message binning (resource killer) suitble for small troubleshooting logs
if you want to analyses some stationary test measurements use time binning

[kahnoie]

I found 1200ms a good enough value for my use.

Regards,
Amit

[seedweb]

hi there,

i've been confronting various types of binning modes in actix lately, and i've seen the results differ quite a bit from mode to mode. Time binning (1000ms) was my default setting, but when i switched to Message binning (1), the results differed 7-8% in each category. Then i tried Time binning (500ms) and the results were similar to message binning. Logic suggests that the smaller the binning time, the more accurate the results, at the expense of processor and memory load. What do you think could be a good trade-off between accuracy and processor load?

Thanks very much

im use classical mode

[kevin_h811]

Hi Serbesi,

Your query is also our query since not all are being plot.

Actix specialist from Denmark, suggested that instead Time binning (set to 1000), we should use Message binning set to 1.

Regards//

[Rooster[BG]]

Hi Guys,

Different binning operation will normally give different results. Keep in mind that histograms and statistics and also "crosstab queries" will give you results, without any binning (message bin). The results on the map are binned (also series in tables), so the easiest way to see different binning results is there.

You should also consider what will happen if you don't pause the measurement and stop for a coffee. If you are close to a site, you will have a lot of samples with very good level, or with very bad level if you are in bad coverage position. In my opinion these result will not be very accurate for the network. I personally use Location binning. This binning gives you geographically distributed results.