This object is in archive! 

GPS accuracy averaging

Rogemar shared this idea 6 years ago
Declined

Can we have an average GPS accuracy in the track summary after the recording for the purpose of knowing the quality of recorded track? Only if possible.


During the recording, the GPS accuracy is also recorded at the time of recording a point based on defined profile (walking, car, etc.). Then, at the end of recording, the gps accuracy associated for each point will be averaged; as a result, it will show in the summary or statistics of the recorded track.

Replies (4)

photo
1

What is the measurement you wish to use to present "GPS accuracy"?

If you export the track in GPX format, each trackpoint's PDOP (Percent Dilution of Precision) is shown.


https://en.wikipedia.org/wiki/Dilution_of_precision_(navigation)

  • Excellent: 1-2
  • Good: 2-5
  • Moderate: 5-10
  • Fair: 10-20
  • Poor: > 20

Tracks contains hundreds or thousands of trackpoints. The average PDOP for so many trackpoints is likely to always fall in to the 2-10 range (Good to Moderate). As a result, I'm not convinced an average value will be meaningful.


Perhaps a more useful indictor would be a pie chart, showing the proportions of trackpoints falling into the categories of Excellent, Good, Moderate, Fair, and Poor. If the pie's slices for Fair and Poor are very large, then the quality of the track's data is low.

photo
1

Hello,

agree with Taras. Average value itself isn't good indicator for quality of any measurement. Huge number of values cover possible problems in certain parts of track so most of track will really end somewhere in middle ( similar value based on your device ). Because we may expect here a normal distribution of values, really useful should be a pie chart as Taras mentioned or a single value that define deviation around this arithmetic mean.

This anyway sounds like a really complicated numbers that will have no real usage. As I think about it, I may only imagine some colored text value like "Record accuracy", colored based on combination of average value and it's distribution without exposing computed numbers to user.

photo
1

Hi Taras and Menion,

Thank you for your feedback, I appreciate it. Just thought about it last night when I'm configuring my dashboard which includes the GPS accuracy that changes from time-to-time. Currently, I'm having a gps accuracy range from 1.5 - 4 meters. I have just thought of the fluctuation of gps accuracy on the track which is kind of similar to altitude change for we will know which part of the track had a poor gps accuracy. If this idea is not feasible and practical, I understand. :)


Also in the dashboard, the gps accuracy is a whole number, can we include 1 decimal? In the Garmin Glo, it provides an accuracy with decimals but in the dashboard it's rounded off. For example, garmin feed is 1.5m while in locus it's 2m.

photo
1

If you don't mind, I will rename your idea to something like "Accuracy indicator of recorded track". I can imagine to set a few groups that will by number and color, define some global accuracy of recorded track. On second side, these values will mainly for longer tracks really close one to another and result will be almost same for all recorded tracks on a single device.

Hmm as I think about it, I'm sorry, I more and more believe it will be quite a lot of work to setup this correctly and no big usage of a result.

About accuracy .. visible accuracy value is not real accuracy, but more a 66% probability that your location is within certain radius. Accuracy of devices grow, but I still believe that for hike & bike activities, rounded number is good enough for quick overview.

Thanks for understanding.

Replies have been locked on this page!