private void PlotLearningCurve(Testing.CurvePoint[] learningCurve, out GraphForm graphFormS, out GraphForm graphFormFn, out GraphForm graphFormFp, double[] simpleAlgorithm, bool excludeLearnedFromTestingSet)
{
// Create 3 new graphs
GraphForm.GraphPlot graphSuccess = new GraphForm.GraphPlot("TrainingSet", "Success");
GraphForm.GraphPlot graphFN = new GraphForm.GraphPlot("TrainingSet", "False Negative");
GraphForm.GraphPlot graphFP = new GraphForm.GraphPlot("TrainingSet", "False Positive");
graphSuccess.setUsedForLearningCurve(true);
graphFN.setUsedForLearningCurve(true);
graphFP.setUsedForLearningCurve(true);
// add point (0,0)
graphSuccess.addPoint(0, 0, "", Color.Black);
graphFN.addPoint(0, 0, "", Color.Black);
graphFP.addPoint(0, 0, "", Color.Black);
foreach (Testing.CurvePoint p in learningCurve)
{
graphSuccess.addPoint(p.getTrainingSetPercentage(), p.getAvgPercentageCorrect(), Math.Round(p.getStdPercentageCorrect(), 2).ToString(), Color.Black);
graphFN.addPoint(p.getTrainingSetPercentage(), p.getAvgPercentageFalseNegative(), Math.Round(p.getStdPercentageFalseNegative(), 2).ToString(), Color.Black);
graphFP.addPoint(p.getTrainingSetPercentage(), p.getAvgPercentageFalsePositive(), Math.Round(p.getStdPercentageFalsePositive(), 2).ToString(), Color.Black);
}
// add point (100,100) or (100,0)
graphSuccess.addPoint(100, 100, "", Color.Black);
graphFN.addPoint(100, 0, "", Color.Black);
graphFP.addPoint(100, 0, "", Color.Black);
//add comparison line for statistical simple algorithms
string partially = (excludeLearnedFromTestingSet) ? "" : "partially_statistical";
graphSuccess.addComparisonLines(simpleAlgorithm[0], "statistical", partially);
graphFN.addComparisonLines(simpleAlgorithm[1], "statistical", partially);
graphFP.addComparisonLines(simpleAlgorithm[2], "statistical", partially);
// Draw
graphFormS = new GraphForm(graphSuccess);
graphFormFn = new GraphForm(graphFN);
graphFormFp = new GraphForm(graphFP);
graphFormS.Show();
graphFormFn.Show();
graphFormFp.Show();
}
private void threadTestParameterClassification()
{
// Get folders
string folderTrue = txtFolderTrueTesting2.Text;
string folderFalse = txtFolderFalseTesting2.Text;
// Get wanted parameters
int parXindex = Convert.ToInt16(txtParamXTesting2.Text);
int parYindex = Convert.ToInt16(txtParamYTesting2.Text);
// Get wanted colors
Color cTrue = pboxColorTrueTesting2.BackColor;
Color cFalse = pboxColorFalseTesting2.BackColor;
// Construct graph
GraphForm.GraphPlot graph = Testing.makeGraph(folderTrue, folderFalse, parXindex, parYindex, cTrue, cFalse);
_graph = graph;
}
/* Making classification graphs for given photos */
public static GraphForm.GraphPlot makeGraph(string folderTruePath, string folderFalsePath, int parX, int parY, Color cTrue, Color cFalse)
{
_progress = 0;
// Get list of files in the folderer (assuming all are images)
string[] filesTrue = LoadImages(folderTruePath);
string[] filesFalse = LoadImages(folderFalsePath);
int numOfFiles = filesTrue.Length + filesFalse.Length;
int filesProcessed = 0;
// Build dictionary for wanted parameters
Dictionary<ImageVector.ImageParameters, bool> parameters = new Dictionary<ImageVector.ImageParameters, bool>();
parameters.Add(ImageVector.getParameterNameByIndex(parX), true);
parameters.Add(ImageVector.getParameterNameByIndex(parY), true);
for (int i = 0; i < ImageVector.NUMBER_OF_PARAMETERS; i++)
if (!parameters.ContainsKey(ImageVector.getParameterNameByIndex(i)))
parameters.Add(ImageVector.getParameterNameByIndex(i), false);
// Convert images to vectors
List<ImageVector> trueList = new List<ImageVector>();
List<ImageVector> falseList = new List<ImageVector>();
foreach (string picPath in filesTrue)
{
trueList.Add(new ImageVector(picPath, parameters));
filesProcessed++;
Progress = (int)((filesProcessed / numOfFiles) * 100);
}
foreach (string picPath in filesFalse)
{
falseList.Add(new ImageVector(picPath, parameters));
filesProcessed++;
Progress = (int)((filesProcessed / numOfFiles) * 100);
}
// Construct graph
string par1name = ImageVector.getParameterNameByIndex(parX).ToString();
string par2name = ImageVector.getParameterNameByIndex(parY).ToString();
GraphForm.GraphPlot graph = new GraphForm.GraphPlot(par1name, par2name);
// Add points to graph
double valueX, valueY;
for (int i = 0; i < trueList.Count; i++)
{
valueX = trueList[i].getParameterByIndex(parX);
valueY = trueList[i].getParameterByIndex(parY);
graph.addPoint(valueX, valueY, i.ToString(), cTrue);
}
for (int i = 0; i < falseList.Count; i++)
{
valueX = falseList[i].getParameterByIndex(parX);
valueY = falseList[i].getParameterByIndex(parY);
graph.addPoint(valueX, valueY, i.ToString(), cFalse);
}
// Get classifications from classifier class
double[] boundsX = new double[0];
double[] boundsY = new double[0];
Classifier.getBoundArray(ImageVector.getParameterNameByIndex(parX), ref boundsX);
Classifier.getBoundArray(ImageVector.getParameterNameByIndex(parY), ref boundsY);
// Convert bounds array to list object and create classification representing char lists
List<double> bX = new List<double>();
List<double> bY = new List<double>();
List<string> bXnames = new List<string>();
List<string> bYnames = new List<string>();
int chr_a = 97;
for (int i = 0; i < boundsX.Length; i++)
{
bX.Add(boundsX[i]);
bXnames.Add(Convert.ToChar(chr_a + i).ToString());
}
bXnames.Add(Convert.ToChar(chr_a + boundsX.Length).ToString());
for (int i = 0; i < boundsY.Length; i++)
{
bY.Add(boundsY[i]);
bYnames.Add(Convert.ToChar(chr_a + i).ToString());
}
bYnames.Add(Convert.ToChar(chr_a + boundsY.Length).ToString());
// Add classification to graph
graph.addXClassification(bX, bXnames);
graph.addYClassification(bY, bYnames);
return graph;
}
