Portfolio Mgmt using Scalding

gistfile1.scala

import com.twitter.scalding._

import cern.colt.matrix.{DoubleFactory2D, DoubleFactory1D }
import cern.colt.matrix.linalg.Algebra
import java.util.StringTokenizer

class Portfolios(args : Args) extends Job(args) {

val cash = 1000.0 // money at hand
val error = 1 // its ok if we cannot invest the last dollar
val (kr,abt,dltr,mnst) = (27.0,64.0,41.0,52.0) // share prices
val stocks = IndexedSeq( kr,abt,dltr,mnst)
weightedSum( stocks, cash,error).write( Tsv("invest.txt"))

// define the correlation matrix
val data = Array(Array(0.448, 0.177, 0.0, 0.017), Array(0.177, 0.393, 0.177, 0.237), Array(0.0, 0.177, 0.237, 0.06), Array(0.017, 0.237, 0.06, 0.19))
val corr = DoubleFactory2D.dense.make(data)

val alg = Algebra.DEFAULT

// read the file into memory
val file = scala.io.Source.fromFile("invest.txt").getLines.toList

// convert the tab-delimited weights in the file to a row vector
def getWeights(s:String) = {  
     val weights = s.split("\t").map( x=> x.toDouble)
     DoubleFactory1D.dense.make(weights)
}

// compute risks per tuple and sort by risk
file.map(line=> { 
     val w = getWeights(line)
     ( line, alg.mult( alg.mult(corr,w), w))
}).sortBy(x=>x._2)

}