[Hpr] Download Stats Question
ken at fallon.ie
Thu Oct 6 12:55:25 PDT 2016
On 2016-10-06 21:13, lostnbronx wrote:
>> OK Let me explain. Should I count all the hits or just downloads ? The
>> feeds or the audio files ? From every ip or filter the bots etc.
>> So basically what would you grep the logs for a HPR download ?
> I would think that each of those, parsed separately, would be of some
> value. It would allow you to see where the most action is coming
From my point of view I could care less how many shows are downloaded.
To me HPR is about the impact of the shows and not the stats. We cannot
judge that except by people giving feedback (which you should do).
However I have lost that argument already before on this list and so
there is an open ticket to add a "number of downloads" count on each
show. There has also been a question "how many subscribers have you"
Each time I started to tackle this topic, I hit a snag in what is a fair
and accurate way to count downloads and subscribers. It's not as simple
as just count the downloads. So for example do we:
First ignore the attackers
Then ignore the bots
The ignore ....
Then count ....
but only if they play download ...% of the file
[do|don't] count duplicates ... in a .... period
(but do count them them if their user agent is different)
Do we filter out shows that have a referral from a kazaa or other music
mp3 sharing sites
etc etc etc.
As a bonus question, how do we deal with the large gaps in our log files
over the years ?
So if you have a think about this and then come back to me with your
algorithm to work this out. I am going to do a show where I run each of
the algorithms and discuss possible pros and cons of each.
I know it would be easier if I could share the file but that would be a
privacy violation and no it can not be anonymised.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 819 bytes
Desc: OpenPGP digital signature
More information about the Hpr