private void kMean(int nodeNumber, int classNumber, double[,] nodes, double[,] classes) // K-mean Algorithm
        {
            //
            // calculate the distance between the class and the coordinate .
            //
            int classNum = classes.GetLength(1);
            double[] point1 = new double[classes.GetLength(1)];
            double[] point2 = new double[classes.GetLength(1)];
            for (int user = 0; user < nodeNumber; user++)
            {
                double min = double.MaxValue;
                for (int attribute = 0; attribute < classes.GetLength(1); attribute++)
                {
                    point1[attribute] = nodes[user, attribute];
                }
                    for (int classsesTag = 0; classsesTag < classNumber; classsesTag++)
                    {
                        for (int classesAttribute = 0; classesAttribute < classes.GetLength(1); classesAttribute++)
                        {
                            point2[classesAttribute] = classes[classsesTag, classesAttribute];
                        }
                        double mindDistance = EuclideanDistance(point1,point2);
                        if (mindDistance < min)
                        {
                            min = mindDistance;
                            nodes[user,classNum] = classsesTag; // flag to set the group that the coordinate belong to.
                        }
                    }
            }
            //
            // calculate the new center class
            //
            double [,] tempClasses = new double[nodeNumber,nodes.GetLength(1)];
            for (int j = 0; j < classNumber; j++)
            {
                double [] tempCoordinate = new double[classes.GetLength(1)];
                for (int i = 0; i < nodeNumber; i++)
                {
                    if (nodes[i,classNum] == j)
                    {
                        for (int tempNodes = 0; tempNodes < classes.GetLength(1); tempNodes++)
                        {
                            tempCoordinate[tempNodes] += nodes[i,tempNodes];
                        }
                        tempClasses[j, classes.GetLength(1)]++;
                    }
                }
                if (tempClasses[j, classes.GetLength(1)] == 0)
                    tempClasses[j, classes.GetLength(1)] = 1;
                for (int newCentral = 0; newCentral < classes.GetLength(1); newCentral++)
                {
                    tempClasses[j, newCentral] = tempCoordinate[newCentral] / tempClasses[j, nodes.GetLength(1)-1];
                }
            }
            //System.Console.ReadKey();
            int k = 0;
            for (k = 0; k < classNumber; k++)
            {
                for (int checkEqual = 0; checkEqual < tempClasses.GetLength(1); checkEqual++)
                {
                    if (tempClasses[k, checkEqual] != classes[k, checkEqual])
                    {
                        recursiveFlag = 1;
                        goto checkBreak;
                     }
                 }
            }
checkBreak: if (k >= classNumber)
                recursiveFlag = 0;
            if (recursiveFlag == 1)
            {
                for (int j = 0; j < classNumber; j++)
                {
                    for (int newClasses = 0; newClasses < classes.GetLength(1); newClasses++)
                    {
                        classes[j, newClasses] = tempClasses[j, newClasses];
                    }
                }
                // recursive
                kMean(nodeNumber, classNumber, nodes, classes);
            }
            if (recursiveFlag == 0)
            {
                //System.Console.WriteLine("=================================");
                //System.Console.WriteLine("finish! recusive times : {0}",loopCount);
                inserCSV(nodes, "clusterFinish");
                inserCSV(classes, "classesCentral");
            }
        }