Activity graphs for the forrest-dev mailing list
4 views
Skip to first unread message
Ramón Casero Cañas
Aug 7, 2008, 7:00:19 AM8/7/08
to Massiel Mailing List
Sent by Israel Herraiz to the OSS Watch mailing list (08/05/2008 04:22 PM):
Hi all,
in the last days I have been working on obtaining the time series for
the activity of the people who participate in the Forrest project.
So far, I have obtained some sparklines that show the activity for the
list and for each one of the people in that list. Have a look at them
at [1].
I am now trying to obtain some metric to find out whether the
individual series have any correlation with the overall activity. In
other words, if any of the individuals is a motivator (or any other
kind of "trigger") that fosters activity at the global level.
I am still trying to figure out how to do that. I will let you know as
soon as I have some "numbers".
Cheers,
Israel
[1] http://gsyc.es/~herraiz/osswatch/graphs/
--
Ramón Casero Cañas
http://www.robots.ox.ac.uk/~rcasero/wiki
http://www.robots.ox.ac.uk/~rcasero/blog
rcasero
Aug 7, 2008, 7:02:12 AM8/7/08
to massiel-talk
Sent by Israel Herraiz to the OSS Watch mailing list (08/05/2008 04:23
PM):
By the way, the information shown in the graphs is number of messages
each week. The vertical bars correspond to one year. The left part is
January 2002. The right part is December 2008.
Cheers,
Israel
PM):
By the way, the information shown in the graphs is number of messages
each week. The vertical bars correspond to one year. The left part is
January 2002. The right part is December 2008.
Cheers,
Israel
rcasero
Aug 7, 2008, 7:03:06 AM8/7/08
to massiel-talk
Sent by Ross Gardler to the OSS Watch mailing list (08/05/2008 09:54
PM):
Israel Herraiz wrote:
> By the way, the information shown in the graphs is number of messages
> each week. The vertical bars correspond to one year. The left part is
> January 2002. The right part is December 2008.
Which lists? (I'd focus on dev to start with, user behaviour is
different)
I'm not sure how these graphs show anything other than posting
activity which I can get easily enough from
http://markmail.org/search/?q=list%3Aorg.apache.forrest.dev#query:list%3Aorg.apache.forrest.dev%20from%3A%22Ross%20Gardler%22+page:1+state:facets
Of course you have the raw data (not sure if markmail provides an API)
In fact, they don't show posting activity very well as the scales are
not consistent.
My understanding was that you were looking for patterns in a poster
creating activity on the lists. Wouldn't that be better studied by
looking at threads rather than overall activity?
There might be 100 posts to a list, but if 90 of those are in a single
thread then clearly one of those threads is more important than the
others.
Take a look at http://people.apache.org/~stefano/agora/ for a tool
that does this kind of work. What is interesting is that you can use
your graphs to look at what you percieve to be changes in the make up
of the team and then use this tool to see if it shows the same
results.
Without a time scale in your latest graphs it is impossible to do it
from there, however, I think you said the big change was the middle of
2003, select a month at the start of 2003 in the above tool and
compare the results to a month at the end of 2003. There is a marked
difference in the cluster of developers shown, then go to the end of
2004 and you see another big difference.
(NB the data and scripts used in this tool are available under the
AL2)
Anyway, I must be missing something about how you are approaching the
problem, I just thought I'd throw the above tool into the mix now as
it may provide another angle - then again, it may be a new path and
therefore should be ignored.
Ross
PM):
Israel Herraiz wrote:
> By the way, the information shown in the graphs is number of messages
> each week. The vertical bars correspond to one year. The left part is
> January 2002. The right part is December 2008.
different)
I'm not sure how these graphs show anything other than posting
activity which I can get easily enough from
http://markmail.org/search/?q=list%3Aorg.apache.forrest.dev#query:list%3Aorg.apache.forrest.dev%20from%3A%22Ross%20Gardler%22+page:1+state:facets
Of course you have the raw data (not sure if markmail provides an API)
In fact, they don't show posting activity very well as the scales are
not consistent.
My understanding was that you were looking for patterns in a poster
creating activity on the lists. Wouldn't that be better studied by
looking at threads rather than overall activity?
There might be 100 posts to a list, but if 90 of those are in a single
thread then clearly one of those threads is more important than the
others.
Take a look at http://people.apache.org/~stefano/agora/ for a tool
that does this kind of work. What is interesting is that you can use
your graphs to look at what you percieve to be changes in the make up
of the team and then use this tool to see if it shows the same
results.
Without a time scale in your latest graphs it is impossible to do it
from there, however, I think you said the big change was the middle of
2003, select a month at the start of 2003 in the above tool and
compare the results to a month at the end of 2003. There is a marked
difference in the cluster of developers shown, then go to the end of
2004 and you see another big difference.
(NB the data and scripts used in this tool are available under the
AL2)
Anyway, I must be missing something about how you are approaching the
problem, I just thought I'd throw the above tool into the mix now as
it may provide another angle - then again, it may be a new path and
therefore should be ignored.
Ross
rcasero
Aug 7, 2008, 7:04:20 AM8/7/08
to massiel-talk
Sent by Israel Herraiz to the OSS Watch mailing list (08/06/2008 10:46
AM):
Excerpts from ross.gardler's message on Aug 5, 2008 about 9 PM:
list.
> > I'm not sure how these graphs show anything other than posting activity
> > which I can get easily enough from
> > http://markmail.org/search/?q=list%3Aorg.apache.forrest.dev#query:list%3Aorg.apa
> > che.forrest.dev%20from%3A%22Ross%20Gardler%22+page:1+state:facets
> >
> > Of course you have the raw data (not sure if markmail provides an API)
Well. I plotted the graphs just for the sake of testing how the data
looks like. But what I am using for the cross-correlation is of course
the raw data.
> > In fact, they don't show posting activity very well as the scales are
> > not consistent.
> >
> > My understanding was that you were looking for patterns in a poster
> > creating activity on the lists. Wouldn't that be better studied by
> > looking at threads rather than overall activity?
Well, I have dropped the scales on purpose. I am just looking at
patterns. Actually, the cross correlation values will be the same
regardless the scale of the series. The only thing that matters is the
scale. I think that is the best option for the kind of things I am
looking for. Anyway, I can always plot the same things with scale.
> > Take a look at http://people.apache.org/~stefano/agora/ for a tool that
> > does this kind of work. What is interesting is that you can use your
> > graphs to look at what you percieve to be changes in the make up of the
> > team and then use this tool to see if it shows the same results.
Thanks. Yes. I can use that as test for the correctness of my data.
> > Anyway, I must be missing something about how you are approaching the
> > problem, I just thought I'd throw the above tool into the mix now as it
> > may provide another angle - then again, it may be a new path and
> > therefore should be ignored.
Well, I admit that I am not being very verbose in my work :-) . As
soon
as I get the cross correlations done, I will start to write a first
draft of the paper, with introduction and methodology sections. I hope
everything will be much clearer then.
Cheers,
Israel
AM):
Excerpts from ross.gardler's message on Aug 5, 2008 about 9 PM:
> > Which lists? (I'd focus on dev to start with, user behaviour is
> > different)
Forrest-dev, with the "filters" that we commented previously in this
> > different)
list.
> > I'm not sure how these graphs show anything other than posting activity
> > which I can get easily enough from
> > http://markmail.org/search/?q=list%3Aorg.apache.forrest.dev#query:list%3Aorg.apa
> > che.forrest.dev%20from%3A%22Ross%20Gardler%22+page:1+state:facets
> >
> > Of course you have the raw data (not sure if markmail provides an API)
looks like. But what I am using for the cross-correlation is of course
the raw data.
> > In fact, they don't show posting activity very well as the scales are
> > not consistent.
> >
> > My understanding was that you were looking for patterns in a poster
> > creating activity on the lists. Wouldn't that be better studied by
> > looking at threads rather than overall activity?
patterns. Actually, the cross correlation values will be the same
regardless the scale of the series. The only thing that matters is the
scale. I think that is the best option for the kind of things I am
looking for. Anyway, I can always plot the same things with scale.
> > Take a look at http://people.apache.org/~stefano/agora/ for a tool that
> > does this kind of work. What is interesting is that you can use your
> > graphs to look at what you percieve to be changes in the make up of the
> > team and then use this tool to see if it shows the same results.
> > Anyway, I must be missing something about how you are approaching the
> > problem, I just thought I'd throw the above tool into the mix now as it
> > may provide another angle - then again, it may be a new path and
> > therefore should be ignored.
soon
as I get the cross correlations done, I will start to write a first
draft of the paper, with introduction and methodology sections. I hope
everything will be much clearer then.
Cheers,
Israel
rcasero
Aug 7, 2008, 7:05:21 AM8/7/08
to massiel-talk
Sent by Ramón Casero to the OSS Watch mailing list (08/06/2008 03:45
PM):
Should we move the discussion about this stuff to the Massiel talk
mailing list?
It's a publicly readable list, so it will be more useful doing it this
way.
Ross, would you like to be subscribed?
Cheers,
.::r
PM):
Should we move the discussion about this stuff to the Massiel talk
mailing list?
It's a publicly readable list, so it will be more useful doing it this
way.
Ross, would you like to be subscribed?
Cheers,
.::r
rcasero
Aug 7, 2008, 7:07:32 AM8/7/08
to massiel-talk
Sent by Pablo Barrera to the OSS Watch mailing list (08/06/2008 04:04
PM):
El 06/08/2008, a las 15:45:31, Ramón Casero Cañas escribió:
>
> Should we move the discussion about this stuff to the Massiel talk mailing list?
>
> It's a publicly readable list, so it will be more useful doing it this way.
>
> Ross, would you like to be subscribed?
What is the idea behind massiel project? What are you trying to do? I
need some background.
Pablo
PM):
El 06/08/2008, a las 15:45:31, Ramón Casero Cañas escribió:
>
> Should we move the discussion about this stuff to the Massiel talk mailing list?
>
> It's a publicly readable list, so it will be more useful doing it this way.
>
> Ross, would you like to be subscribed?
need some background.
Pablo
rcasero
Aug 7, 2008, 7:08:35 AM8/7/08
to massiel-talk
Sent by Ramón Casero to the OSS Watch mailing list (08/06/2008 04:09
PM):
Pablo Barrera wrote:
>
> El 06/08/2008, a las 15:45:31, Ramón Casero Cañas escribió:
>
>>
>> Should we move the discussion about this stuff to the Massiel talk mailing list?
>>
>> It's a publicly readable list, so it will be more useful doing it this way.
>>
>> Ross, would you like to be subscribed?
>
>
> What is the idea behind massiel project? What are you trying to do? I need some background.
Massiel is a Google Code project where Israel put his scripts to do
analysis of open source communities.
http://code.google.com/p/massiel
Ideally, we would like to find people who are interested in this kind
of research and join the project.
The mailing list is the right place for scientific as well as dev
discussion, I think.
Cheers,
.::r
PM):
Pablo Barrera wrote:
>
> El 06/08/2008, a las 15:45:31, Ramón Casero Cañas escribió:
>
>>
>> Should we move the discussion about this stuff to the Massiel talk mailing list?
>>
>> It's a publicly readable list, so it will be more useful doing it this way.
>>
>> Ross, would you like to be subscribed?
>
>
> What is the idea behind massiel project? What are you trying to do? I need some background.
analysis of open source communities.
http://code.google.com/p/massiel
Ideally, we would like to find people who are interested in this kind
of research and join the project.
The mailing list is the right place for scientific as well as dev
discussion, I think.
Cheers,
.::r
rcasero
Aug 7, 2008, 7:13:13 AM8/7/08
to massiel-talk
Sent by Pablo Barrera to the OSS Watch mailing list (08/06/2008 04:15
PM):
El 06/08/2008, a las 16:09:15, Ramón Casero Cañas escribió:
> Pablo Barrera wrote:
>> El 06/08/2008, a las 15:45:31, Ramón Casero Cañas escribió:
>>>
>>> Should we move the discussion about this stuff to the Massiel talk mailing list?
>>>
>>> It's a publicly readable list, so it will be more useful doing it this way.
>>>
>>> Ross, would you like to be subscribed?
>> What is the idea behind massiel project? What are you trying to do? I need some background.
>
>
> Massiel is a Google Code project where Israel put his scripts to do analysis of open source communities.
>
> http://code.google.com/p/massiel
>
> Ideally, we would like to find people who are interested in this kind of research and join the project.
>
Do a bunch of scripts need a google code project? I still don't see
where are you trying to do. Israel scripts are too problem specific to
have a useful goal outside his next paper. If you have a broader goal
in mind, what is it?
> The mailing list is the right place for scientific as well as dev discussion, I think.
>
Then maybe be should move this conversation there.
Pablo
PM):
El 06/08/2008, a las 16:09:15, Ramón Casero Cañas escribió:
> Pablo Barrera wrote:
>> El 06/08/2008, a las 15:45:31, Ramón Casero Cañas escribió:
>>>
>>> Should we move the discussion about this stuff to the Massiel talk mailing list?
>>>
>>> It's a publicly readable list, so it will be more useful doing it this way.
>>>
>>> Ross, would you like to be subscribed?
>> What is the idea behind massiel project? What are you trying to do? I need some background.
>
>
> Massiel is a Google Code project where Israel put his scripts to do analysis of open source communities.
>
> http://code.google.com/p/massiel
>
> Ideally, we would like to find people who are interested in this kind of research and join the project.
>
where are you trying to do. Israel scripts are too problem specific to
have a useful goal outside his next paper. If you have a broader goal
in mind, what is it?
> The mailing list is the right place for scientific as well as dev discussion, I think.
>
Pablo
Reply all
Reply to author
Forward
0 new messages