Median is defined as the middle value of a sorted list, or if the list
has an even number of elements, the average of the two middle values of
the sorted list.
The code below uses this definition to calculate the median. If this
code gives the same answer as commons math for your large dataset, then
perhaps Excel is wrong.
double median(double[] values) {
if (values.length == 0) {
return Double.NaN; // act same as commons.Math
}
double[] copy = new double[values.length];
System.arraycopy(values, 0, copy, 0, values.length);
java.util.Arrays.sort(copy);
int midpos = copy.length / 2;
if (copy.length % 2 == 1) { // odd number of elements
return copy[midpos]; // return middle element
}
// even number of elements: return avg of two middle elements
return (copy[midpos - 1] + copy[midpos]) / 2;
}
I tested with some small datasets, and this method gave the same answer
as commons.Math...
Hope this helps,
--Remko Popma
Jeff Drew wrote:
I'm having a weird problem when using the commons math package. When I run
statistics using the Commons math, then compare the results to Excel, I get
different standard deviation and median, but min, max, and count are the
same. I'd appreciate any ideas on how Commons Math and Excel differ in
these calculations.
MEDIAN: Excel: 468,231 CommonsMath: 485,711
STD: Excel: 11,861 CommonsMath: 10,678
The data set is 18,000 integers so I won't include those. They are
mostly 6
digit numbers. Here's the code:
import org.apache.commons.math.stat.descriptive.moment.StandardDeviation;
import org.apache.commons.math.stat.descriptive.rank.Max;
import org.apache.commons.math.stat.descriptive.rank.Median;
import org.apache.commons.math.stat.descriptive.rank.Min;
import gnu.trove.TDoubleHashSet;
public class ExampleForMailingList {
StandardDeviation std = new StandardDeviation( );
Min min = new Min( );
Max max = new Max( );
Median medianInstance = new Median();
private double minimum = 0;
private double maximum = 0;
private double standardDev = 0;
private double median = 0;
private boolean isCalcDone = false;
private double count = 0;
/**
* <code>data</code> If the length is zero, then only 0 measurements
were added.
*/
TDoubleHashSet data = new TDoubleHashSet( );
/**
* If the <code>measurement</code> is greater than 0, then add it to the
data.
*
* @param measurement
*/
public void addMeasurement( int measurement ) {
data.add( measurement );
count++;
}
/**
* Must be called before using the getters. This method calculates the
statistics.
*/
public void calculate() {
try {
double[] dataArray = data.toArray( );
minimum = min.evaluate( dataArray );
maximum = max.evaluate( dataArray );
standardDev = std.evaluate( dataArray );
median = medianInstance.evaluate(dataArray);
isCalcDone = true;
} catch ( RuntimeException e ) {
// TODO Auto-generated catch block
e.printStackTrace( );
}
} // calculate
public double getMinimum() throws CalcNotDoneException {
return minimum;
} // get minimum
public double getMaximum() throws CalcNotDoneException {
return maximum;
} // get maximum
public double getStd() throws CalcNotDoneException {
return standardDev;
} // get std
public double getMedian() throws CalcNotDoneException {
return median;
} // get median
/**
* Converts a result set into a set of statistics which a table model
consumes. Calculates: <br>
* 1. min <br>
* 2. average <br>
* 3. max <br>
* 4. median<br>
* 5. percent threshold violations <br>
* @param resultSetArg
* Results of an order table query
*/
public void processResults( ResultSet results,String column ) {
int value = Integer.MAX_VALUE;
try {
while ( results.next( ) ) {
value = ( int ) results.getLong( column );
if ( value > -1 ) {
addMeasurement( value );
}
}
} catch ( SQLException e ) {
// TODO Auto-generated catch block
e.printStackTrace();
} // while
} // processResults
public static void main( String[] args ) {
ExampleForMailingList example = new ExampleForMailingList();
example.processResults(ResultSet set,"columnA");
example.calculate( );
System.out.println("std: "+ example.getStd( ));
System.out.println("std: "+ example.getMedian( ));
}
}
Thanks!
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
