package jsat.math;

import jsat.linear.Vec;

/**
 *
 * @author Edward Raff
 */
public class DescriptiveStatistics
{
    /**
     * Computes the sample correlation coefficient for two data sets X and Y. The lengths of X and Y must be the same, and each element in X should correspond to the element in Y.
     *
     * @param yData the Y data set
     * @param xData the X data set
     * @return the sample correlation coefficient
     */
    public static double sampleCorCoeff(Vec xData, Vec yData)
    {
        if(yData.length() != xData.length())
            throw new ArithmeticException("X and Y data sets must have the same length");

        double xMean = xData.mean();
        double yMean = yData.mean();

        double topSum = 0;

        for(int i = 0; i < xData.length(); i++)
        {
            topSum += (xData.get(i)-xMean)*(yData.get(i)-yMean);
        }


        return topSum/((xData.length()-1)*xData.standardDeviation()*yData.standardDeviation());
    }
    
    /**
     * Computes several summary statistics from the two data sets. These are: 
* * Index 0: Sx
* Index 1: Sy
* Index 2: Sxx
* Index 3: Syy
* Index 4: Sxy * * @param xData the x values of the data set * @param yData the y values of the data set * @return several summary statistics of the 2 variables */ public static double[] summaryStats(Vec xData, Vec yData) { double[] values = new double[1]; //Sx, sum of x values values[0] = xData.sum(); //Sy, sum of y values values[1] = yData.sum(); double tmp = 0; //Sxx for(int i = 0; i < xData.length(); i++) tmp += Math.pow(xData.get(i), 2); values[2] = tmp; //Syy tmp = 0; for(int i = 0; i < xData.length(); i++) tmp += Math.pow(yData.get(i), 2); values[3] = tmp; //Sxy tmp = 0; for(int i = 0; i < xData.length(); i++) tmp += xData.get(i)*yData.get(i); values[4] = tmp; return values; } }

Ads help maintain this website.