-1

what is the step-by-step procedure for executing a program in mahout

Anakha
  • 435
  • 1
  • 5
  • 4

4 Answers4

3

Maybe you can check the MAHOUT wiki? Especially the quickstart section looks interesting...

Fortega
  • 19,463
  • 14
  • 75
  • 113
  • ya we saw the document..since we are really new to maout, we are not able to approach it in the right manner..could you give us a step-by-step answer if you have worked with it? – Anakha Jan 22 '10 at 13:24
  • I'm sorry, than I cannot help you. I once wanted to start working with it, so I remembered reading parts of this wiki. But I never really found the time to really start with it... – Fortega Jan 22 '10 at 14:31
3

Basically what worked from me was following their wiki: https://cwiki.apache.org/MAHOUT/buildingmahout.html

  1. svn co http://svn.apache.org/repos/asf/mahout/trunk
  2. cd trunk
  3. mvn install
  4. cd core
  5. mvn compile
  6. mvn install
  7. cd examples
  8. mvn compile

Then follow what to do with the examples, for example the taste-web example, you need to download the data set, load it in the war, and edit which recommender system to use.

Mahout is awesome, give it a try, and you will enjoy it :)

Mohamed Mansour
  • 39,445
  • 10
  • 116
  • 90
1

Mahout isn't an execution environment, so I'm not sure what you mean by 'run a program in Mahout?' It is a library, and one that does lots of things: recommendation engines, classification, clustering, more. You haven't given any detail about what you're doing.

I am a dev of Mahout and you could get more useful answers by being more specific and asking at mahout-user@apache.org.

Also, are you looking at the javadoc at all? The Mahout Hadoop-related jobs all document their arguments.

Sean Owen
  • 66,182
  • 23
  • 141
  • 173
  • Sean, the documentation to mahout is pitifully minimal. The major short comings are how to load data from files into to the system and how to format the data in those files. A chapter of your book could be devoted, or even an appendix could be devoted to file format. Clustering does have some of this explanation but it focuses heavily on loading your data then writing it to a SequenceFile to be loaded again into a Vector and not really explaining the logic behind this. Unless a coder works close with the development team, they will have a hard time understanding the software architecture. – Adam Hess May 29 '12 at 18:42
  • I don't really disagree, but open source is open source: it's up to everyone, not only me, to document it as well as create it. I think the recommender bits are actually pretty well documented. The file format is, simply, "CSV", which is explained in the javadoc and the book. I don't know what you're doing, or your question, so don't know how to respond really. – Sean Owen May 29 '12 at 22:12
0

http://girlincomputerscience.blogspot.com.br/2010/11/apache-mahout.html

jen
  • 194
  • 1
  • 3