This modules uses my gpxcsv module to convert to gpx files to pandas dataframes and perform simple analysis of speed, pace, and time. You can see it in action via my web app.

git clone https://github.com/astrowonk/gpxrun.git
cd gpxrun
python setup.py install

from gpxrun import GpxRun
g = GpxRun('myfile.gpx.gz')
#summary frame of 1 row
g.summary_data

#gpxcsv data frame augmented with computed rows
g.gpx_data

Example image of an apparently painfully slow run:

Also includes a helper function gpx_multi which can process a glob of files and return a concattenated dataframe summary.

df = gpx_multi('*.gpx')

At least when it comes to the Apple Watch, pace/distance information shown in the Fitness app or during a workout is based on the pedometer, not the GPS. In theory, the Watch calibrates itself using the GPS so the pedometer is accurate. I know that I had disabled location services for Motion Calibration and Distance, and that when I finaly turned this on, suddenly the Watch started calibrating and within a run or two, it was telling me I was slower than I had been.

In practice with gpxrun, an uncalibrated Watch will have the largest discrepencies with the GPS-based pace/speed/distance, but this gap will close if/as it is calibrated. So far, in about a week of calibration, the self-reported distances from the Watch pedometer are about 2%-3% larger than what I find in the GPS data in the GPX exported files (using HealthFit).

How accurate is the pedometer can be when fully calibrated, I do not know yet. I will eventually update this readme comparing pace/distance data from gpxrun with what the pedometer-based distance iOS reports.

Another question, how accurate is GPS based pace/distance? The Watch reports GPS accuracy and it's +/- a meter or two. Presumably over a long run, the total integrated distance is fairly accurate.