Principal Component Analysis with Breeze (Scala)

pca.sc

// For an unkown reason, this works when copy/paste in ammonite but not with `amm pca.sc`
import $ivy.`org.scalanlp::breeze-natives:0.13.2`
import $ivy.`org.scalanlp::breeze-viz:0.13.2`
import $ivy.`org.scalanlp::breeze:0.13.2`

import breeze.linalg._
import breeze.linalg.svd._
import breeze.plot._

import scala.util.Random._


val dimensions = 50
val values = 200

def mean(v: Vector[Double]): Double = v.valuesIterator.sum / v.size

def zeroMean(m: DenseMatrix[Double]): DenseMatrix[Double] = {
  val copy = m.copy
  for (c <- 0 until m.cols) {
    val col = copy(::, c)
    val colMean = mean(col)
    col -= colMean
  }
  //    println("data \n" + m)
  //    println("mean \n" + copy)
  copy
}

def pca(data: DenseMatrix[Double], components: Int): DenseMatrix[Double] = {
  val d = zeroMean(data)
  val SVD(_, _, v) = svd(d.t)
  val model = v(0 until components, ::) //top 'components' eigenvectors
  val filter = model.t * model
  filter * d
}

//val data= csvread(new File("data.csv"), skipLines = 1).t
//  val data = DenseMatrix(
//    (2.0, 4.0, 5.1),
//    (1.0, 2.5, 3.5),
//    (8.0, 3.0, 6.4),
//    (8.0, 5.0, 6.5),
//    (4.3, 4.5, 6.4))

def generateData: DenseMatrix[Double] = {
  val data = DenseMatrix.zeros[Double](values, dimensions)
  for (d <- 0 until dimensions) {
    val c1 = 2 + 2 * (nextDouble - 0.5)
    val r1 = nextDouble * 2
    val c2 = -2 + 2 * (nextDouble - 0.5)
    val r2 = nextDouble * 2
    for (v <- 0 until values / 2) {
      val vv = c1 + r1 * nextDouble
      //      println(f"a:$a b:$b d:$d v:$v vv:$vv%2f")
      data.update(v, d, vv)
    }
    for (v <- values / 2 until values) {
      val vv = c2 + r2 * nextDouble
      //      println(f"a:$a b:$b d:$d v:$v vv:$vv%2f")
      data.update(v, d, vv)
    }
  }
  data
}

val data = generateData

val pcaRes = pca(data, 2)
// or better since https://github.com/scalanlp/breeze/commit/913e1229cb3572b43062ee1f756858ac793bb8b8#diff-64378cd6c871f77715faf60c46568a8e
//  val pca = princomp(data)
//  val pcaRes = pca.scores
println("result pca \n" + pcaRes)

val f1 = Figure("data")
val f2 = Figure("pca")
f1.subplot(0) += scatter(data(::, 0), data(::, 3), { _ => 0.1 })
f2.subplot(0) += scatter(pcaRes(::, 0), pcaRes(::, 1), { _ => 0.1 })

As per Breeze 1.0-RC2, line 22 should be:

val SVD(_, _, v) = svd(d.t)

@frgomes thanks ! updated with ammonite