You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by Stefano Bellasio <st...@gmail.com> on 2010/11/12 09:55:36 UTC
Creating a Sparse Matrix to use with SVD
HHi guys, i'm new with Mahout and Hadoop. I tried something and right now i did some tests with recommendation (based on grouplens dataset) with Mahout and with Hadoop; i'm trying to figure out how i can use SVD for recommendations (Lanczos solver distributed with Hadoop), and i know that i need a matrix to start. So i coded something in Java using SparseDoubleMatrix2D (in the math dir of mahout) but i'm not sure of what i have to do.
My code:
package org.apache.mahout.math.matrix.impl;
import org.apache.mahout.math.function.BinaryFunction;
public class creaMatrice extends SparseDoubleMatrix2D {
public static void main(String args[]){
double array[][]= { {1,2,4,8}, {5,0,4,5}, {2,9,1,5} };
SparseDoubleMatrix2D matrice = new SparseDoubleMatrix2D(array);
int prova = matrice.getQuick(2,3);
System.out.println(prova);
}
}
It's just to try if something works and the answer is "no" for now :) But firstly i have some questions about SVD, matrix and hadoop.
1) i need to create a matrix with item-ratings and then convert it with SVD application of mahout?
2) is there a guide or something else that can explain this?
Thank you for your support, hope to see some helpful answer :) Stefano