How to calculate the average of dgp-s values?

I would like to compare glare of my models;

Model a: A room with one window and with louvers

Model b: A room with one window and with no louvers

According to output of OS (glare.json), I have a large number of dgp-s. I think that to compare glare of my models, I have to average all dgp-s values (and I guess there is no need to raw values).

If I am not wrong my question is that how can I calculate the average of all dgp-s? (as they are 8760 values and calculating by hand is really time consuming). image description

Yes, of course you have a large number of values. You have one for each hour of the year, for each sensor point (view). Your screenshot is showing some data for 10:00 a.m., and we can see six views from a single glare sensor, for each model. The first value is for the primary view (the one the arrow points to in the GUI), and the rest are arrayed equally about the sensor's location. So, lots of values.

These DGP(s) values are highly dependent upon view direction, and the glare results file includes (in your case) multiple view directions; averaging will smear the metric, likely in ways you don't want. I recommend looking at Jakubiec's paper about the "Adaptive Zone" for more details.

See my comments on your other question as well. I think you need to gain an understanding of what this glare metric is all about, in order to decide what to average (if anything), and what it is you want to say with your data.