示例#1
0
 public Signal()
 {
     pca      = new PCA();
     knn      = new KNN();
     svm      = new SVM();
     svmscale = new SVMScale();
 }
示例#2
0
        public Tuple <double[], double[], double[], double[]> getDistance(KNN obj, double[] test1, double[] test2)
        {
            double[] distancedown  = obj.EuclideanDistance(downT7, downT8, test1, test2);
            double[] distanceup    = obj.EuclideanDistance(upT7, upT8, test1, test2);
            double[] distanceleft  = obj.EuclideanDistance(leftT7, leftT8, test1, test2);
            double[] distanceright = obj.EuclideanDistance(rightT7, rightT8, test1, test2);

            return(new Tuple <double[], double[], double[], double[]>(distancedown, distanceup, distanceleft, distanceright));
        }
示例#3
0
        public void buildCorpus(string subName = "karrar")
        {
            KNN    P            = new KNN();
            string channelName1 = "IED_T8";

            downT8  = P.readCSV(subName + "Down", channelName1);
            upT8    = P.readCSV(subName + "Up", channelName1);
            rightT8 = P.readCSV(subName + "Right", channelName1);
            leftT8  = P.readCSV(subName + "Left", channelName1);

            string channelName2 = "IED_T7";

            downT7  = P.readCSV(subName + "Down", channelName2);
            upT7    = P.readCSV(subName + "Up", channelName2);
            rightT7 = P.readCSV(subName + "Right", channelName2);
            leftT7  = P.readCSV(subName + "Left", channelName2);
        }
        public double calculateKNNAccuracy()
        {
            if (fileExistance)
            {
                KNN knn = new KNN();

                if (rightTestData != null)
                {
                    knn.setRight(rightTrainData);
                }

                if (leftTestData != null)
                {
                    knn.setLeft(leftTrainData);
                }

                if (upTestData != null)
                {
                    knn.setUp(upTrainData);
                }

                if (downTestData != null)
                {
                    knn.setDown(downTrainData);
                }

                if (neutralTestData != null)
                {
                    knn.setNeutral(neutralTrainData);
                }

                double tp = 0, fp = 0;
                if (rightTestData != null)
                {
                    for (int i = 0; i < rightTestData.Length; i++)
                    {
                        if (knn.knnClassifier(rightTestData[i]) == "Right")
                        {
                            tp++;
                        }
                        else
                        {
                            fp++;
                        }
                    }
                }
                if (leftTestData != null)
                {
                    for (int i = 0; i < leftTestData.Length; i++)
                    {
                        if (knn.knnClassifier(leftTestData[i]) == "Left")
                        {
                            tp++;
                        }
                        else
                        {
                            fp++;
                        }
                    }
                }
                if (upTestData != null)
                {
                    for (int i = 0; i < upTestData.Length; i++)
                    {
                        if (knn.knnClassifier(upTestData[i]) == "Up")
                        {
                            tp++;
                        }
                        else
                        {
                            fp++;
                        }
                    }
                }
                if (downTestData != null)
                {
                    for (int i = 0; i < downTestData.Length; i++)
                    {
                        if (knn.knnClassifier(downTestData[i]) == "Down")
                        {
                            tp++;
                        }
                        else
                        {
                            fp++;
                        }
                    }
                }
                if (neutralTestData != null)
                {
                    for (int i = 0; i < neutralTestData.Length; i++)
                    {
                        if (knn.knnClassifier(neutralTestData[i]) == "Neutral")
                        {
                            tp++;
                        }
                        else
                        {
                            fp++;
                        }
                    }
                }
                double accuracy = (tp / (tp + fp)) * 100;
                return(accuracy);
            }
            else
            {
                return(0);
            }
        }