private void btn_KNN(object sender, RoutedEventArgs e)
{
int checkresutl = 0;
var lp = int.Parse(txt_LenghtOfPattern.Text);
var kNN = int.Parse(txt_KNN.Text);
var rate = float.Parse(txt_rateIP.Text);
var numPre = int.Parse(txt_NumOfPredic.Text);
var dataPrediction = txt_SelectedData.Text;
checkresutl = CheckDataLenghPattern(lp);
checkresutl = CheckValueKNN(kNN);
checkresutl = CheckRateImportantPoint(rate);
checkresutl = CheckPredictionHorizon(numPre);
checkresutl = CheckDataInput(dataPrediction);
if (checkresutl.Equals(0))
{
List <DataPrediction> dataNeedPre = new List <DataPrediction>();
List <DataPrediction> dataAfterPre = new List <DataPrediction>();
List <DataPrediction> actualData = new List <DataPrediction>();
KNearestNeighbor KNN = new KNearestNeighbor(lp, kNN, rate);
//actualData = Function.ReadCSVFile(@"F:\TTTN_LVTN\Duong Tuan Anh\SourceCode\LVTNDTA\Data\LEADING_ECONOMIC.csv");
dataNeedPre = Function.ReadCSVFile(dataPrediction);
dataAfterPre = KNN.PredictWithKNN(dataNeedPre);
for (int i = 1; i < numPre; i++)
{
Function.RefeshData(dataAfterPre);
dataAfterPre = KNN.PredictWithKNN(dataAfterPre);
}
//Function.MAE(actualData, dataAfterPre);
LineChart chart = new LineChart(dataAfterPre);
chart.Show();
}
}
public KnnBot(Checker myColor) : base(myColor)
{
learner = new KNearestNeighbor(5, 1000);
// this learner needs to have at least one example
// to start with
Example e = MakeExample(new Board(), Checker.Blue);
e.Labels.Add(0.5);
LearnOneExample(e);
}
public void Can_classify_with_euclidean_distance()
{
string[] labels;
var featureSpace = GetDataSet(out labels);
var b = KNearestNeighbor.Classify(new Vector(0, 0), featureSpace, labels, 3);
Assert.AreEqual("B", b);
var a = KNearestNeighbor.Classify(new Vector(1.0, 1.0), featureSpace, labels, 3);
Assert.AreEqual("A", a);
}
private void button3_Click(object sender, EventArgs e)
{
Glass newGlass = new Glass(-1, (double)riNum.Value, (double)naNum.Value, (double)mgNum.Value, (double)alNum.Value, (double)siNum.Value, (double)kNum.Value, (double)caNum.Value, (double)baNum.Value, (double)feNum.Value, -1);
var neuralOutput = network.GetOutputs(newGlass);
output(("[BACK PROPOGATION] Siūloma tipas: "******"[K-Nearest-Neighbors] Siūloma tipas: "******"[Bayes] Siūloma tipas: " + bayes.test(newGlass)));
}
private object FindPreviousTransactionByPayee(Account a, Transaction t, string payeeOrTransferCaption)
{
MyMoney money = t.MyMoney;
IList <Transaction> list = money.Transactions.GetTransactionsFrom(a);
int len = list.Count;
if (len == 0)
{
// Nothing to do here
return(null);
}
// Tally of how close the current transaction is to the amounts for a given category or split.
singleNeighbors = new KNearestNeighbor <Category>();
splitNeighbors = new KNearestNeighbor <Category>();
splitCount = normalCount = 0;
Transaction closestByDate = null;
long ticks = 0;
decimal amount = t.Amount;
for (int i = 0; i < len; i++)
{
Transaction u = list[i] as Transaction;
if (amount == 0)
{
// we can't use the probabilities when the amount is zero, so we just return
// the closest transaction by date because in the case of something like a paycheck
// the most recent paycheck usually has the closest numbers on the splits.
long newTicks = Math.Abs((u.Date - t.Date).Ticks);
if (closestByDate == null || newTicks < ticks)
{
closestByDate = u;
ticks = newTicks;
}
}
else
{
AddPossibility(t, u, payeeOrTransferCaption);
}
}
if (closestByDate != null)
{
return(closestByDate);
}
IEnumerable <Tuple <object, Category> > result = null;
if (splitCount > normalCount)
{
result = splitNeighbors.GetNearestNeighbors(1, t.Amount);
}
else
{
result = singleNeighbors.GetNearestNeighbors(1, t.Amount);
}
if (result != null && result.Any())
{
var first = result.First();
var it = first.Item1 as Transaction;
if (it != null && it.IsSplit)
{
closestByDate = null;
ticks = 0;
// if this is a "Split" transaction, then we should grab the closest date
// so that the copied splits have the best chance of matching.
// (e.g. in a split paycheck scenario)
foreach (var u in list)
{
if (u.IsSplit && u.PayeeOrTransferCaption == t.PayeeOrTransferCaption)
{
long newTicks = Math.Abs((u.Date - t.Date).Ticks);
if (closestByDate == null || newTicks < ticks)
{
closestByDate = u;
ticks = newTicks;
}
}
}
return(closestByDate);
}
return(first.Item1);
}
return(null);
}
public void KNearestNeighborConstructorTest()
{
double[][] inputs =
{
new double[] { -5, -2, -1 },
new double[] { -5, -5, -6 },
new double[] { 2, 1, 1 },
new double[] { 1, 1, 2 },
new double[] { 1, 2, 2 },
new double[] { 3, 1, 2 },
new double[] { 11, 5, 4 },
new double[] { 15, 5, 6 },
new double[] { 10, 5, 6 },
};
int[] outputs =
{
0, 0,
1, 1, 1, 1,
2, 2, 2
};
int k = 3;
KNearestNeighbor target = new KNearestNeighbor(k, inputs, outputs);
for (int i = 0; i < inputs.Length; i++)
{
int actual = target.Compute(inputs[i]);
int expected = outputs[i];
Assert.AreEqual(expected, actual);
}
double[][] test =
{
new double[] { -4, -3, -1 },
new double[] { -5, -4, -4 },
new double[] { 5, 3, 4 },
new double[] { 3, 1, 6 },
new double[] { 10, 5, 4 },
new double[] { 13, 4, 5 },
};
int[] expectedOutputs =
{
0, 0,
1, 1,
2, 2,
};
for (int i = 0; i < test.Length; i++)
{
int actual = target.Compute(test[i]);
int expected = expectedOutputs[i];
Assert.AreEqual(expected, actual);
}
}
private void Knn_start_Click(object sender, EventArgs e)
{
double averageAccurasy = 0;
List <Glass> glasses = new List <Glass>();
if (radioButton1.Checked)
{
glasses = dataList;
}
else
{
glasses = clearDistantValuesFromList(dataList);
}
int testingFileCount = glasses.Count / NUMBER_OF_SEGMENTS;
for (int segment = 0; segment < NUMBER_OF_SEGMENTS; segment++)
{
richTextBox1.AppendText(String.Format("Crosscheck segment {0} \n ", segment + 1));
int correctGuesses = 0;
int guessCounter = 0;
int falseGuesses = 0;
int testingDataFromIndex = segment * testingFileCount;
int testingDataToIndex = (segment + 1) * testingFileCount;
var trainingData = new List <Glass>();
var testingData = new List <Glass>();
for (int i = 0; i < glasses.Count; i++)
{
if (i >= testingDataFromIndex && i <= testingDataToIndex)
{
testingData.Add(glasses[i]);
}
else
{
trainingData.Add(glasses[i]);
}
}
var knn = new KNearestNeighbor(7, trainingData);
foreach (Glass item in testingData)
{
var ans = knn.classify(item);
guessCounter++;
if (ans == item.group_type)
{
correctGuesses++;
}
else
{
falseGuesses++;
}
}
var totalGlasses = glasses.Count;
var totalTrain = trainingData.Count;
var totalTest = testingData.Count;
double trainProc = Math.Round(((double)totalTrain / totalGlasses) * 100, 2);
double testProc = Math.Round(((double)totalTest / totalGlasses) * 100, 2);
double accuracy = Math.Round(((double)correctGuesses / guessCounter) * 100, 2);
averageAccurasy += accuracy;
richTextBox1.AppendText(string.Format("Total: {0, 3}; Training: {1, 3}({2, 4}%); Testing: {3, 3}({4, 4}%)\n", totalGlasses, totalTrain, trainProc, totalTest, testProc));
richTextBox1.AppendText(string.Format("Correct guesses: {0, 3} ; Total guesses: {1, 3} ; Accuracy: {2, 4}%\n", correctGuesses, guessCounter, accuracy));
}
richTextBox1.AppendText(string.Format("Average accurasy of all segments - {0}%\n", averageAccurasy / NUMBER_OF_SEGMENTS));
}
public void KNearestNeighborConstructorTest()
{
double[][] inputs =
{
new double[] { -5, -2, -1 },
new double[] { -5, -5, -6 },
new double[] { 2, 1, 1 },
new double[] { 1, 1, 2 },
new double[] { 1, 2, 2 },
new double[] { 3, 1, 2 },
new double[] { 11, 5, 4 },
new double[] { 15, 5, 6 },
new double[] { 10, 5, 6 },
};
int[] outputs =
{
0, 0,
1, 1, 1, 1,
2, 2, 2
};
int k = 3;
KNearestNeighbor target = new KNearestNeighbor(k, inputs, outputs);
for (int i = 0; i < inputs.Length; i++)
{
int actual = target.Compute(inputs[i]);
int expected = outputs[i];
Assert.AreEqual(expected, actual);
}
double[][] test =
{
new double[] { -4, -3, -1 },
new double[] { -5, -4, -4 },
new double[] { 5, 3, 4 },
new double[] { 3, 1, 6 },
new double[] { 10, 5, 4 },
new double[] { 13, 4, 5 },
};
int[] expectedOutputs =
{
0, 0,
1, 1,
2, 2,
};
for (int i = 0; i < test.Length; i++)
{
int actual = target.Compute(test[i]);
int expected = expectedOutputs[i];
Assert.AreEqual(expected, actual);
}
}
