mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paritosh Ranjan <pran...@xebia.com>
Subject Re: Why compiling Mahout from trunk if there are already builds available in maven public repo?
Date Tue, 18 Sep 2012 08:36:24 GMT
You can also use Mahout Command Line Invocation to run algorithms. i.e. 
see the .sh file, its has Mahout commands inside which can be used 
individually.

'With complicated I mean (1) getting the source from svn-trunk, (2)
compiling the source with maven, (3) generating an eclipse-maven project
with "mvn eclipse:eclipse"?'

Though this is not mahout related, but you can also give a try to maven 
scm connector plugin for eclipse. It will automatically checkout the 
code as a maven project, which will save few steps.

On 18-09-2012 08:21, Dhruv wrote:
> The approach you should follow depends on what you want to do.
>
> If you're trying to use a stable Mahout release for your app, use approach
> #2 which you have listed below.
>
> if you're trying to extend Mahout, use the bleeding edge features or peek
> into its development, then you should download trunk, compile it etc.
>
> Running examples using the shell scripts is a way to understand the
> algorithms, no matter what you want to do with the library.
>
> On Mon, Sep 17, 2012 at 3:13 PM, <sysrq@web.de> wrote:
>
>> Hi,
>>
>> why every tutorial I found so far makes integrating Eclipse with Mahout so
>> complicated?
>> With complicated I mean (1) getting the source from svn-trunk, (2)
>> compiling the source with maven, (3) generating an eclipse-maven project
>> with "mvn eclipse:eclipse"?
>>
>> I am just starting to use Mahout and want to run as many examples as I can
>> to get experience about using clustering algorithms "the right way".
>>
>> If I understood correctly, then there are two ways to interact with Mahout
>> and write "programs" for it:
>>   1. with scripts in bin/ folder, which are boundled into a single batch.sh
>> script file
>>   2. programming against the Mahout Java API, so "embedding Mahout" in a
>> Java application
>>
>> With which approach should I start?
>> I am trying to start with approach 2., but maybe I should start 1.? I
>> don't know.
>>
>> My way to start with approach 2. is very simple
>>   1. Create a new Maven-project in Eclipse with File > New > Other > Maven
>> Project
>>      So no need to deal with CLI or "mvn eclipce:eclipse" at all
>>   2. In Eclipse double-click on pom.xml, open the "Dependendencies" tab,
>> click on "Add...",
>>      search for "mahout" and select "mahout-core", which basically includes
>> the following:
>>        <dependency>
>>          <groupId>org.apache.mahout</groupId>
>>          <artifactId>mahout-core</artifactId>
>>          <version>0.7</version>
>>          </dependency>
>>        </dependencies>
>>
>> This way is simple and straightforward. But why then is every tutorial
>> talking about the "hard way" with svn-trunk, and compiling and so on? Or am
>> I missing somethign important here?
>>



Mime
View raw message