public string Predict(List <double> marks)
{
try
{
if (marks.Count() < 10)
{
return("غير معلوم");
}
if (Model == null)
{
if (!File.Exists("randomforest.xml"))
{
Learn("data.txt", 500, 266, 10);
}
Model = ClassificationForestModel.Load(() => new StreamReader(@"randomforest.xml"));
}
var lastTenMarks = marks.Skip(marks.Count - 10);
F64Matrix observations = new F64Matrix(1, lastTenMarks.Count());
for (var j = 0; j < lastTenMarks.Count(); j++)
{
observations.At(0, j, lastTenMarks.ElementAt(j));
}
var v = Model.Predict(observations).First();
Console.WriteLine("predict " + v);
return(GetClass(v));
}
catch (Exception e)
{
Console.Error.WriteLine(e.Message);
return("غير معلوم");
}
}
/// <summary>
/// Converts a list of n LearningCurvePoint to a n by 3 matrix
/// with columns SampleSize, TrainingScore and ValidationScore.
/// </summary>
/// <param name="points"></param>
/// <returns></returns>
public static F64Matrix ToF64Matrix(this List <LearningCurvePoint> points)
{
if (points.Count == 0)
{
throw new ArgumentException("There must be at least one element in the list to convert to a matrix");
}
;
var matrix = new F64Matrix(points.Count, 3);
for (int i = 0; i < points.Count; i++)
{
var point = points[i];
matrix.At(i, 0, point.SampleSize);
matrix.At(i, 1, point.TrainingScore);
matrix.At(i, 2, point.ValidationScore);
}
return(matrix);
}
/// <summary>
/// Transposes matrix.
/// Output is saved in the provided matrix transposed.
/// </summary>
/// <param name="matrix"></param>
/// <param name="transposed"></param>
/// <returns></returns>
public static void TransposeF64(F64Matrix matrix, F64Matrix transposed)
{
var cols = matrix.ColumnCount;
var rows = matrix.RowCount;
for (int i = 0; i < rows; i++)
{
for (int j = 0; j < cols; j++)
{
transposed.At(j, i, matrix.At(i, j));
}
}
}
/// <summary>
///
/// </summary>
/// <param name="ensemblePredictions"></param>
/// <param name="predictions"></param>
public void Combine(F64Matrix ensemblePredictions, double[] predictions)
{
var cols = ensemblePredictions.ColumnCount;
var rows = ensemblePredictions.RowCount;
for (int i = 0; i < rows; i++)
{
var sum = 0.0;
for (int j = 0; j < cols; j++)
{
sum += ensemblePredictions.At(i, j);
}
predictions[i] = sum / (double)cols;
}
}
static void Parse(string dataPath, F64Matrix observations, List <double> targets, F64Matrix testObservations, List <double> actualTargets, int FEATURES_COUNT, int LEARNING_ROWS_COUNT)
{
var lines = File.ReadAllLines(dataPath);
int TOTAL_LINES = lines.Length;
int i = 0;
foreach (var line in lines)
{
var j = 0;
var values = line.Trim().Split(' ');
if (values.Length > FEATURES_COUNT)
{
foreach (var t in values)
{
if (j >= LEARNING_ROWS_COUNT)
{
actualTargets.Add(Normlize(Double.Parse(t)));
}
else
{
targets.Add(Normlize(Double.Parse(t)));
}
j++;
}
}
else
{
foreach (var r in values)
{
if (i < LEARNING_ROWS_COUNT)
{
observations.At(i, j, Normlize(Double.Parse(r)));
}
else
{
testObservations.At(i - LEARNING_ROWS_COUNT, j, Normlize(Double.Parse(r)));
}
j++;
}
}
i++;
}
}
/// <summary>
/// Converts F64Matrix to float[][].
/// </summary>
/// <param name="matrix"></param>
/// <returns></returns>
public static float[][] ToFloatJaggedArray(this F64Matrix matrix)
{
var rows = matrix.RowCount;
var cols = matrix.ColumnCount;
var jaggedArray = new float[rows][];
for (int row = 0; row < rows; row++)
{
var rowArray = new float[cols];
for (int col = 0; col < cols; col++)
{
rowArray[col] = (float)matrix.At(row, col);
}
jaggedArray[row] = rowArray;
}
return(jaggedArray);
}
/// <summary>
/// Converts F64Matrix to float[][].
/// </summary>
/// <param name="matrix"></param>
/// <param name="rowIndices"></param>
/// <returns></returns>
public static float[][] ToFloatJaggedArray(this F64Matrix matrix, int[] rowIndices)
{
var rows = rowIndices.Length;
var cols = matrix.ColumnCount;
var jaggedArray = new float[rows][];
for (int outputRow = 0; outputRow < rowIndices.Length; outputRow++)
{
var inputRow = rowIndices[outputRow];
var rowArray = new float[cols];
for (int col = 0; col < cols; col++)
{
rowArray[col] = (float)matrix.At(inputRow, col);
}
jaggedArray[outputRow] = rowArray;
}
return(jaggedArray);
}
void FindBestSplit(F64Matrix observations, double[] residuals, double[] targets, double[] predictions, int[][] orderedElements,
GBMTreeCreationItem parentItem, bool[] parentInSample, int featureIndex, ConcurrentBag <GBMSplitResult> results)
{
var bestSplit = new GBMSplit
{
Depth = parentItem.Depth,
FeatureIndex = -1,
SplitIndex = -1,
SplitValue = -1,
Cost = double.MaxValue,
LeftConstant = -1,
RightConstant = -1,
SampleCount = parentItem.Values.Samples
};
var bestLeft = GBMSplitInfo.NewEmpty();
var bestRight = GBMSplitInfo.NewEmpty();
var left = GBMSplitInfo.NewEmpty();
var right = parentItem.Values.Copy(NodePositionType.Right);
var orderedIndices = orderedElements[featureIndex];
var j = NextAllowedIndex(0, orderedIndices, parentInSample);
// No allowed sample or valid split left.
if (j >= orderedIndices.Length || orderedIndices.Length == 1)
{
return;
}
var currentIndex = orderedIndices[j];
m_loss.UpdateSplitConstants(ref left, ref right, targets[currentIndex], residuals[currentIndex]);
var previousValue = observations.At(currentIndex, featureIndex);
while (right.Samples > 0)
{
j = NextAllowedIndex(j + 1, orderedIndices, parentInSample);
currentIndex = orderedIndices[j];
var currentValue = observations.At(currentIndex, featureIndex);
if (Math.Min(left.Samples, right.Samples) >= m_minimumSplitSize)
{
if (previousValue != currentValue)
{
var cost = left.Cost + right.Cost;
if (cost < bestSplit.Cost)
{
bestSplit.FeatureIndex = featureIndex;
bestSplit.SplitIndex = j;
bestSplit.SplitValue = (previousValue + currentValue) * .5;
bestSplit.LeftError = left.Cost;
bestSplit.RightError = right.Cost;
bestSplit.Cost = cost;
bestSplit.CostImprovement = parentItem.Values.Cost - cost;
bestSplit.LeftConstant = left.BestConstant;
bestSplit.RightConstant = right.BestConstant;
bestLeft = left.Copy();
bestRight = right.Copy();
}
}
}
m_loss.UpdateSplitConstants(ref left, ref right, targets[currentIndex], residuals[currentIndex]);
previousValue = currentValue;
}
if (bestSplit.FeatureIndex != -1)
{
results.Add(new GBMSplitResult {
BestSplit = bestSplit, Left = bestLeft, Right = bestRight
});
}
}
/// <summary>
/// Returns the agumented version of the data. Excluding the original.
/// The each feature in the dataset must be scaled/normnalized between 0.0 and 1.0
/// before the method works.
/// </summary>
/// <param name="dataset"></param>
/// <returns></returns>
public F64Matrix Agument(F64Matrix dataset)
{
var orgCols = dataset.ColumnCount;
var orgRows = dataset.RowCount;
var augmentation = new F64Matrix(dataset.RowCount, dataset.ColumnCount);
var indicesVisited = new HashSet <int>();
var sample = new double[orgCols];
var candidate = new double[orgCols];
indicesVisited.Clear();
for (int j = 0; j < orgRows; j++)
{
if (indicesVisited.Contains(j))
{
continue;
}
dataset.Row(j, sample);
var closestDistance = double.MaxValue;
var closestIndex = -1;
indicesVisited.Add(j);
for (int f = 0; f < orgRows; f++)
{
if (indicesVisited.Contains(f))
{
continue;
}
dataset.Row(f, candidate);
var distance = GetDistance(sample, candidate);
if (distance < closestDistance)
{
closestDistance = distance;
closestIndex = f;
}
}
if (closestIndex != -1)
{
dataset.Row(closestIndex, candidate);
indicesVisited.Add(closestIndex);
for (int h = 0; h < sample.Length; h++)
{
var sampleValue = sample[h];
var candiateValue = candidate[h];
if (m_random.NextDouble() <= m_probabilityParameter && m_probabilityParameter != 0.0)
{
var std = (sampleValue - candiateValue) / m_localVariance;
augmentation.At(j, h, SampleRandom(candiateValue, std));
augmentation.At(closestIndex, h, SampleRandom(sampleValue, std));
}
else // keep values
{
augmentation.At(j, h, sampleValue);
augmentation.At(closestIndex, h, candiateValue);
}
}
}
}
return(augmentation);
}