Example usage for weka.core Statistics normalProbability

Introduction

In this page you can find the example usage for weka.core Statistics normalProbability.

Prototype

public static double normalProbability(double a) 

Source Link

Document

Returns the area under the Normal (Gaussian) probability density function, integrated from minus infinity to x (assumes mean is zero, variance is one).

Usage

From source file:moa.classifiers.core.driftdetection.STEPD.java

License:Open Source License

@Override
public void input(double prediction) { // In MOA, 1.0=false, 0.0=true.
    if (!this.isInitialized) {
        initialize();/*from   ww  w .j  ava2  s.  c om*/
        this.isInitialized = true;
    } else {
        if (this.isChangeDetected) {
            resetLearning();
        }
    }

    if (nr == windowSize) { // Recent window is full.
        wo = wo + storedPredictions[firstPos]; // Oldest prediction in recent window 
        no++; // is moved to older window,
        wr = wr - storedPredictions[firstPos];
        firstPos++; // Start of recent window moves.
        if (firstPos == windowSize) {
            firstPos = 0;
        }
    } else { // Recent window grows.
        nr++;
    }

    lastPos++; // Adds prediction at the end of recent window.
    if (lastPos == windowSize) {
        lastPos = 0;
    }
    ;
    storedPredictions[lastPos] = (byte) prediction;
    wr += prediction;

    this.isWarningZone = false;

    if (no >= windowSize) { // The same as: (no + nr) >= 2 * windowSize.
        ro = no - wo; // Numbers of correct predictions are calculated.
        rr = nr - wr;
        sizeInvertedSum = 1.0 / no + 1.0 / nr; // Auxiliary variable.
        p = (ro + rr) / (no + nr); // Calculation of the statistics of STEPD.
        Z = Math.abs(ro / no - rr / nr);
        Z = Z - sizeInvertedSum / 2.0;
        Z = Z / Math.sqrt(p * (1.0 - p) * sizeInvertedSum);

        Z = Statistics.normalProbability(Math.abs(Z));
        Z = 2 * (1 - Z);

        if (Z < alphaDrift) { // Drift Level
            this.isChangeDetected = true;
        } else {
            if (Z < alphaWarning) { // Warning Level
                this.isWarningZone = true;
            }
        }
    }
}

---

From source file:tclass.util.FastMath.java

License:Open Source License

/**
 * Returns the normalCDF for the value; accurate to 10^-6 or so. 
 *//*from ww  w .j a  va2 s. c  o m*/
public static float normalCDF(float val) {
    if (val < min) {
        return (float) Statistics.normalProbability(val);
    }
    if (val > max) {
        return (float) Statistics.normalProbability(val);
    } else {
        //Find array position: 
        float scaledVal = (val - min) / interval;
        int arrayIndex = (int) scaledVal;
        float remainder = scaledVal - arrayIndex;
        float lowerBnd = nCDFArray[arrayIndex];
        float upperBnd = nCDFArray[arrayIndex + 1];
        return (lowerBnd + (upperBnd - lowerBnd) * remainder);
    }
}

---

From source file:tclass.util.FastMath.java

License:Open Source License

public static float logNormalCDF(float val) {
    if (val < min) {
        return (float) Math.log(Statistics.normalProbability(val));
    }//from w w w  .  j  a v  a2  s. c  o  m
    if (val > max) {
        return (float) Math.log(Statistics.normalProbability(val));
    } else {
        //Find array position: 
        float scaledVal = (val - min) / interval;
        int arrayIndex = (int) scaledVal;
        float remainder = scaledVal - arrayIndex;
        float lowerBnd = nLCDFArray[arrayIndex];
        float upperBnd = nLCDFArray[arrayIndex + 1];
        return (lowerBnd + (upperBnd - lowerBnd) * remainder);
    }
}

---

From source file:tclass.util.FastMath.java

License:Open Source License

public static void main(String args[]) {
    for (float i = -4.955f; i < 5.1; i += 0.1) {
        System.out.println("In: " + i + " N: " + normalCDF(i) + " LN: " + logNormalCDF(i) + " Ex: "
                + Statistics.normalProbability(i) + " Err: "
                + Math.abs(normalCDF(i) - Statistics.normalProbability(i)));
    }/*w ww  . j  ava 2  s. co  m*/
    float maxErr = 0;
    float maxErrPoint = 0;
    for (float i = -4.995f; i < 5.1; i += 0.01) {
        if (Math.abs(normalCDF(i) - Statistics.normalProbability(i)) > maxErr) {
            maxErr = (float) Math.abs(normalCDF(i) - Statistics.normalProbability(i));
            maxErrPoint = i;
        }
    }
    System.err.println("Max Error Point = " + maxErrPoint + " Max Error = " + maxErr);
}

---

From source file:tucil.Kernelatr.java

@Override
public double getProbability(double data) {
    double delta, sum = 0, currentProb;
    double zLower, zUpper;
    if (numvalue == 0) {
        zLower = (data - (precision / 2)) / stdev;
        zUpper = (data + (precision / 2)) / stdev;
        return (Statistics.normalProbability(zUpper) - Statistics.normalProbability(zLower));
    }//from  w  w w  .  ja  v  a 2s  .c  o m

    double weightSum = 0;
    int start = findNearestValue(data);
    for (int i = start; i < numvalue; i++) {
        delta = values[i] - data;
        zLower = (delta - (precision / 2)) / stdev;
        zUpper = (delta + (precision / 2)) / stdev;
        currentProb = (Statistics.normalProbability(zUpper) - Statistics.normalProbability(zLower));
        sum += currentProb * weights[i];

        weightSum += weights[i];
        if (currentProb * (sumweight - weightSum) < sum * maxerror) {
            break;
        }
    }
    for (int i = start - 1; i >= 0; i--) {
        delta = values[i] - data;
        zLower = (delta - (precision / 2)) / stdev;
        zUpper = (delta + (precision / 2)) / stdev;
        currentProb = (Statistics.normalProbability(zUpper) - Statistics.normalProbability(zLower));
        sum += currentProb * weights[i];
        weightSum += weights[i];
        if (currentProb * (sumweight - weightSum) < sum * maxerror) {
            break;
        }
    }
    return sum / sumweight;
}

---

From source file:tucil.Normalatr.java

@Override
public double getProbability(double data) {
    data = round(data);/*from   w  w  w  .j  a  v a 2 s  . c  o m*/
    double zlower = (data - mean - (precision / 2)) / stdev;
    double zupper = (data - mean + (precision / 2)) / stdev;

    double plower = Statistics.normalProbability(zlower);
    double pupper = Statistics.normalProbability(zupper);
    return pupper - plower;
}

---