Example #1
0
        /// <summary>
        /// Perform an Hierarchical clustering on each plate independantely
        /// </summary>
        /// <param name="CurrentPlateToProcess">the plate to process</param>
        /// <param name="ClassNumber">Number of class</param>
        private void ClusteringHierarchicalSinglePlate(cPlate CurrentPlateToProcess, int ClassNumber)
        {
            weka.core.Instances Ninsts = CurrentPlateToProcess.CreateInstancesWithoutClass();

            weka.clusterers.HierarchicalClusterer HClusterer = new HierarchicalClusterer();

            //string OptionDistance = " -A \"weka.core.";
            string OptionDistance = " -A \"";

            switch (cGlobalInfo.OptionsWindow.comboBoxHierarchicalDistance.SelectedIndex)
            {
                case 0:
                    OptionDistance += "EuclideanDistance";
                    break;
                case 1:
                    OptionDistance += "ManhattanDistance";
                    break;
                case 2:
                    OptionDistance += "ChebyshevDistance";
                    break;
                default:
                    break;
            }

            OptionDistance += " -R first-last\"";

            string[] TAGS_LINK_TYPE = { "SINGLE", "COMPLETE", "AVERAGE", "MEAN", "CENTROID", "WARD", "ADJCOMPLETE" };

            string WekaOption = "-L " + TAGS_LINK_TYPE[cGlobalInfo.OptionsWindow.comboBoxHierarchicalLinkType.SelectedIndex];// + OptionDistance;

            HClusterer.setOptions(weka.core.Utils.splitOptions(WekaOption));
            //EuclideanDistance2 Dist2 = new EuclideanDistance2();

            //HClusterer.setDistanceFunction(Dist2);
            HClusterer.setNumClusters(ClassNumber);
            HClusterer.buildClusterer(Ninsts);

            richTextBoxInfoClustering.AppendText("\n" + CurrentPlateToProcess.GetName() + ": " + HClusterer.numberOfClusters() + " cluster(s)");

            ClusterEvaluation eval = new ClusterEvaluation();
            eval.setClusterer(HClusterer);
            eval.evaluateClusterer(Ninsts);

            CurrentPlateToProcess.AssignClass(eval.getClusterAssignments());
        }
Example #2
0
        /// <summary>
        /// perform an Hierarchical clustering over the entire screening data
        /// </summary>
        /// <param name="ClassNumber"></param>
        private void ClusteringHierarchicalGlobalScreen(int ClassNumber)
        {
            weka.core.Instances Ninsts = CompleteScreening.CreateInstancesWithoutClass();
            weka.clusterers.HierarchicalClusterer HClusterer = new HierarchicalClusterer();

            string OptionDistance = " -A \"weka.core.";

            switch (GlobalInfo.OptionsWindow.comboBoxHierarchicalDistance.SelectedIndex)
            {
                case 0:
                    OptionDistance += "EuclideanDistance";
                    break;
                case 1:
                    OptionDistance += "ManhattanDistance";
                    break;
                case 2:
                    OptionDistance += "ChebyshevDistance";
                    break;
                default:
                    break;
            }

            OptionDistance += " -R first-last\"";

            string[] TAGS_LINK_TYPE = { "SINGLE", "COMPLETE", "AVERAGE", "MEAN", "CENTROID", "WARD", "ADJCOMPLETE" };

            string WekaOption = "-L " + TAGS_LINK_TYPE[GlobalInfo.OptionsWindow.comboBoxHierarchicalLinkType.SelectedIndex] + OptionDistance;

            HClusterer.setOptions(weka.core.Utils.splitOptions(WekaOption));
            HClusterer.setNumClusters(ClassNumber);
            HClusterer.buildClusterer(Ninsts);

            richTextBoxInfoClustering.AppendText("\n" + HClusterer.numberOfClusters() + " cluster(s) identified");

            ClusterEvaluation eval = new ClusterEvaluation();
            eval.setClusterer(HClusterer);
            eval.evaluateClusterer(Ninsts);

            CompleteScreening.AssignClass(eval.getClusterAssignments());
        }