That's too small to be that slow. There are a bunch of ways this could be
slower than it should.
The DataSource may not matter. What's important is whether it is actually a
pooling DataSource from the container. You may want to check whether it
seems to be reusing connections.
Table indexes, on us
Hey,
I got the same problem, of slowness while using MYSQL data model, after a
small research and looking into mysql's query log, revealed that user - user
recommendation just floods the database with thousands and thousands of
requests.
and thats on small database.
for now Im dumbping the database
That's all correct, it reads a lot. But you can avoid a lot of it by using
caching wrappers.
You also don't need to dump to a file. Use ReloadFromJDBCDataModel.
On Wed, Jul 13, 2011 at 9:57 AM, Vitali Mogilevsky
wrote:
> Hey,
> I got the same problem, of slowness while using MYSQL data model, aft
Hi
Because I am a new user, so I will appreciate for a table like this:
Algorithm Name Current Status Local-Run Commands
Map-Reduce Run Commands Dataset file format
LogisticRegression ProductiontrainAdaptiveLogistic... N/A
CSV with head
Thanks, will test that
On Wed, Jul 13, 2011 at 12:11 PM, Sean Owen wrote:
> That's all correct, it reads a lot. But you can avoid a lot of it by using
> caching wrappers.
> You also don't need to dump to a file. Use ReloadFromJDBCDataModel.
>
> On Wed, Jul 13, 2011 at 9:57 AM, Vitali Mogilevsky
I think your table got kind of hashed up.
Can you put your table on a Mahout wiki page?
On Wed, Jul 13, 2011 at 7:11 AM, Xiaobo Gu wrote:
> Hi
>
> Because I am a new user, so I will appreciate for a table like this:
>
> Algorithm Name Current Status Local-Run Commands
> Map-Reduce Run
Awesome, I will give ReloadFromJDBCDataModel a try. How does this
particular data model update itself on database changes? Does it just
happen periodically and if so, can this rate be change easily?
Lastly, will calling clear(userId) on a recommender frequently be bad
for performance? I'm assuming
-Original Message-
From: Salil Apte
Date: Wed, 13 Jul 2011 10:19:47
To:
Reply-To: user@mahout.apache.org
Subject: Re: Connection Pooling
Awesome, I will give ReloadFromJDBCDataModel a try. How does this
particular data model update itself on database changes? Does it just
happen period
Yes it reloads after a configurable interval, or on demand.
Clearing the cache for a user ID only means that user's data is recomputed.
It's not bad to call this frequently per se... I suppose you want to let it
cache as much and for as long as is valid and acceptable to your app.
Your bottleneck
The only objection to this is that it is yet another piece of information to
be maintained, and there is a strong chance it will not quite be kept up to
date. We already have a bit of doc rot in the javadoc itself, and the wiki
(which is just par for the course in a volunteer project).
I would fin
Where can the interval be configured? BTW, ReloadFromJDBCDataModel
works like a dream so far :)
On Wed, Jul 13, 2011 at 10:58 AM, Sean Owen wrote:
> Yes it reloads after a configurable interval, or on demand.
>
> Clearing the cache for a user ID only means that user's data is recomputed.
> It's n
I was mixing this up with another class. It doesn't reload itself. You can
call refresh() to do so.
On Wed, Jul 13, 2011 at 7:34 PM, Salil Apte wrote:
> Where can the interval be configured? BTW, ReloadFromJDBCDataModel
> works like a dream so far :)
>
> On Wed, Jul 13, 2011 at 10:58 AM, Sean Ow
Agreed. And I would encourage the original poster to put their table onto
the wiki and post a JIRA pointing to the classes they would like better
javadoc in.
It is always easier to respond to specifics than make open-ended wide-spread
improvements (Sean is excepted from this generalization).
On
Hello,
I'm looking for more similarity metrics, such as Hellinger distance.
Wouldn't they be implemented as a subclass of DistributedVectorSimilarity?
Does anyone have more implementations?
Thanks, Ian
What's in the project now is all I know about. Yes if you want to use it
with the Hadoop-based similarity calculator, that's what you would extend.
How do you apply this metric to vectors?
On Wed, Jul 13, 2011 at 10:09 PM, Ian Upright wrote:
> Hello,
>
> I'm looking for more similarity metrics,
You would have to encode the distributions as vectors.
For discrete distributions, I think that this is relatively trivial since
you could interpret each vector entry as the probability for an element i of
the domain of the distribution. I think that would result in the Hellinger
distance [1] bei
I found this:
http://www.utdallas.edu/~herve/Abdi-Distance2007-pretty.pdf
Which seems to explain it pretty simply. Seems like these measures should
be fairly easy to implement. I could take a stab at it and publish the
results.
Ian
>What's in the project now is all I know about. Yes if you wa
If you need this distance, please go for it!
The procedure for publishing the results (or the first attempts) is to file
a JIRA (see issues.apache.org/jira/browse/MAHOUT ) and attach patches to the
JIRA for review or comment.
On Wed, Jul 13, 2011 at 2:55 PM, Ian Upright wrote:
> I found this:
>
Yes that's it, according to the reference -- or rather, I suppose, construe
the vector as encoding a discrete distribution. Element i has probability
proportion to the value at i. (It can't have negative values of course.)
It would seem to be what you write below, but the square root of the sum of
19 matches
Mail list logo