[tor-bugs] #21255 [Metrics/Metrics website]: fraction value computation for clients.csv and its descriptions don't seem to match

["Tor Bug Tracker & Wiki" <blackhole@xxxxxxxxxxxxxx>]

#21255: fraction value computation for clients.csv and its descriptions don't seem
to match
-----------------------------------------+--------------------------
     Reporter:  iwakeh                   |      Owner:  metrics-team
         Type:  defect                   |     Status:  new
     Priority:  Medium                   |  Milestone:
    Component:  Metrics/Metrics website  |    Version:
     Severity:  Normal                   |   Keywords:
Actual Points:                           |  Parent ID:
       Points:                           |   Reviewer:
      Sponsor:                           |
-----------------------------------------+--------------------------
 Maybe, I'm missing something obvious, but the calculation for `frac` in
 `clients.csv` doesn't seem to calculate what is stated in the web-site
 description (nor in the sql comment),
 [https://metrics.torproject.org/stats.html#clients web-site]:
   "frac: Fraction of relays or bridges in percent that the estimate is
 based on."


 [https://gitweb.torproject.org/metrics-web.git/tree/modules/clients/init-
 userstats.sql?id=14840ed2db075bbc1d0991b974becc3826a50969#n553 SQL
 excerpt]:
 {{{
 ...
   -- Estimated fraction of nodes reporting directory requests, which is
   -- used to extrapolate observed requests to estimated total requests in
   -- the network.  The closer this fraction is to 1.0, the more precise
   -- the estimation.
   CAST(a.frac * 100 AS INTEGER) AS frac,

   -- Finally, the estimate number of users.
   CAST(a.rrx / (a.frac * 10) AS INTEGER) AS users

   -- Implement the estimation method in a subquery, so that the ugly
   -- formula only has to be written once.
   FROM (
     SELECT date, node, country, transport, version, rrx, nrx,
            (hrh * nh + hh * nrh) / (hh * nn) AS frac
 <--------------------<<<
     FROM aggregated WHERE hh * nn > 0.0) a

   -- Only include estimates with at least 10\% of nodes reporting
 directory
   -- request statistics.
   WHERE a.frac BETWEEN 0.1 AND 1.0
 ...
 }}}

 The arrow points at the **fraction of reported directory requests** (or
 responses for bridges) of the total (estimated) sum of directory requests
 (responses for bridges), but not the **fraction of nodes reporting
 directory requests** of the total number of nodes.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/21255>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs