public void KappaTestConstructorTest2()
{
// Example from: M. Reichenheim (2004). Confidence intervals for the
// kappa statistic. The Stata Journal (2004) 4, Number 4, pp. 421–428.
// http://www.stata-journal.com/sjpdf.html?articlenum=st0076
int[,] matrix = // (pg 599)
{
{ 48, 12 },
{ 16, 160 },
};
GeneralConfusionMatrix a = new GeneralConfusionMatrix(matrix);
// Reichenheim's paper shows:
// Agreement | Expected Agreement | Kappa | Std. Error | Z
// 88.14% | 61.25% | 0.6938 | 0.0650 | 10.67
Assert.AreEqual(88.14, a.OverallAgreement * 100, 1e-2);
Assert.AreEqual(61.25, a.ChanceAgreement * 100, 1e-2);
Assert.AreEqual(0.6938, a.Kappa, 1e-4);
Assert.AreEqual(0.0650, a.StandardErrorUnderNull, 1e-4);
KappaTest target = new KappaTest(a);
Assert.AreEqual(OneSampleHypothesis.ValueIsGreaterThanHypothesis, target.Hypothesis);
Assert.AreEqual(10.67, target.Statistic, 1e-3);
Assert.AreEqual(7.0849733798130419E-27, target.PValue);
}
public void KappaTestConstructorTest4()
{
// Example from Statistical Methods for Rates and Proportions
// Checked against http://graphpad.com/quickcalcs/Kappa2.cfm (OK)
// Checked against Statistical Methods for Rates and Proportions (OK)
// Checked against http://vassarstats.net/kappa.html (FAIL)
int[,] matrix = // (pg 599)
{
{ 75, 1, 4 },
{ 5, 4, 1 },
{ 0, 0, 10 },
};
GeneralConfusionMatrix a = new GeneralConfusionMatrix(matrix);
Assert.AreEqual(100, a.Samples);
Assert.AreEqual(80, a.RowTotals[0]);
Assert.AreEqual(10, a.RowTotals[1]);
Assert.AreEqual(10, a.RowTotals[2]);
Assert.AreEqual(80, a.ColumnTotals[0]);
Assert.AreEqual(5, a.ColumnTotals[1]);
Assert.AreEqual(15, a.ColumnTotals[2]);
double[,] proportions = // (pg 599)
{
{ 0.75, 0.01, 0.04 },
{ 0.05, 0.04, 0.01 },
{ 0.00, 0.00, 0.10 },
};
Assert.IsTrue(proportions.IsEqual(a.ProportionMatrix));
double expectedVar = a.Variance;
// Test against non-null hypothesis
KappaTest target = new KappaTest(a, hypothesizedKappa: 0.8);
// Fleiss reports 0.68 (page 606)
// Graphpad reports 0.676
Assert.AreEqual(0.68, target.EstimatedValue, 0.01);
Assert.AreEqual(a.Kappa, target.EstimatedValue);
Assert.IsFalse(double.IsNaN(target.EstimatedValue));
// Fleiss reports 0.087 (page 607)
// Graphpad reports 0.088
Assert.AreEqual(0.087, target.StandardError, 0.001);
Assert.IsFalse(double.IsNaN(target.StandardError));
Assert.AreEqual(-1.38, target.Statistic, 0.029);
Assert.AreEqual(0.1589, target.PValue, 0.0001);
Assert.IsFalse(target.Significant);
}
public void KappaTestConstructorTest5()
{
// Example from Statistical Methods for Rates and Proportions
// for kappa variance under the null hypothesis
// Checked against Statistical Methods for Rates and Proportions (OK)
int[,] matrix = // (pg 599)
{
{ 75, 1, 4 },
{ 5, 4, 1 },
{ 0, 0, 10 },
};
GeneralConfusionMatrix a = new GeneralConfusionMatrix(matrix);
Assert.AreEqual(100, a.Samples);
Assert.AreEqual(80, a.RowTotals[0]);
Assert.AreEqual(10, a.RowTotals[1]);
Assert.AreEqual(10, a.RowTotals[2]);
Assert.AreEqual(80, a.ColumnTotals[0]);
Assert.AreEqual(5, a.ColumnTotals[1]);
Assert.AreEqual(15, a.ColumnTotals[2]);
double[,] proportions = // (pg 599)
{
{ 0.75, 0.01, 0.04 },
{ 0.05, 0.04, 0.01 },
{ 0.00, 0.00, 0.10 },
};
Assert.IsTrue(proportions.IsEqual(a.ProportionMatrix));
// Test under null hypothesis
KappaTest target = new KappaTest(a, hypothesizedKappa: 0,
alternate: OneSampleHypothesis.ValueIsGreaterThanHypothesis);
Assert.AreEqual(0.68, target.EstimatedValue, 0.01); // pg 605
Assert.AreEqual(a.Kappa, target.EstimatedValue);
Assert.IsFalse(double.IsNaN(target.EstimatedValue));
Assert.AreEqual(0.076, target.StandardError, 0.001);
Assert.IsFalse(double.IsNaN(target.StandardError));
Assert.AreEqual(8.95, target.Statistic, 0.08);
Assert.IsTrue(target.Significant);
}
public void KappaTestConstructorTest()
{
// Example from http://vassarstats.net/kappa.html
// Checked against http://graphpad.com/quickcalcs/Kappa2.cfm (OK)
int[,] matrix =
{
{ 44, 5, 1 },
{ 7, 20, 3 },
{ 9, 5, 6 },
};
GeneralConfusionMatrix a = new GeneralConfusionMatrix(matrix);
Assert.AreEqual(a.RowTotals[0], 50);
Assert.AreEqual(a.RowTotals[1], 30);
Assert.AreEqual(a.RowTotals[2], 20);
Assert.AreEqual(a.ColumnTotals[0], 60);
Assert.AreEqual(a.ColumnTotals[1], 30);
Assert.AreEqual(a.ColumnTotals[2], 10);
Assert.AreEqual(0.4915, a.Kappa, 1e-4);
Assert.IsFalse(double.IsNaN(a.Kappa));
double var = a.Variance;
double var0 = a.VarianceUnderNull;
double varD = Accord.Statistics.Testing.KappaTest.DeltaMethodKappaVariance(a);
double se = System.Math.Sqrt(var);
double se0 = System.Math.Sqrt(var0);
double seD = System.Math.Sqrt(varD);
Assert.AreEqual(0.072, a.StandardError, 0.0005);
// Create a test of the null hypothesis (actual k = 0)
KappaTest target = new KappaTest(a, hypothesizedKappa: 0);
// Std. Error is computed differently under the null hypothesis:
Assert.AreEqual(0.073509316753225237, target.StandardError, 1e-5);
Assert.IsFalse(double.IsNaN(target.StandardError));
}
public void KappaTestConstructorTest3()
{
int[,] matrix =
{
{ 377, 79, 0, 0 },
{ 2, 72, 0, 0 },
{ 33, 5, 60, 0 },
{ 3, 20, 0, 8 },
};
GeneralConfusionMatrix a = new GeneralConfusionMatrix(matrix);
Assert.AreEqual(0.7845, a.OverralAgreement, 1e-4);
Assert.IsFalse(double.IsNaN(a.OverralAgreement));
Assert.AreEqual(0.47986, a.ChanceAgreement, 1e-5);
Assert.IsFalse(double.IsNaN(a.ChanceAgreement));
Assert.AreEqual(0.586, a.Kappa, 1e-3);
Assert.IsFalse(double.IsNaN(a.Kappa));
KappaTest target = new KappaTest(a);
Assert.AreEqual(0.586, target.Kappa, 1e-3);
Assert.AreEqual(a.Kappa, target.Kappa);
Assert.IsFalse(double.IsNaN(target.Kappa));
Assert.AreEqual(0.00087457, target.Variance, 1e-5);
Assert.IsFalse(double.IsNaN(target.Variance));
Assert.AreEqual(19.806, target.Statistic, 0.1);
Assert.IsFalse(double.IsNaN(target.Statistic));
Assert.AreEqual(0.644, target.Confidence.Max, 1e-3);
Assert.AreEqual(0.528, target.Confidence.Min, 1e-3);
Assert.AreEqual(0.0, target.PValue, 1e-6);
Assert.IsFalse(double.IsNaN(target.PValue));
Assert.IsTrue(target.Significant);
}
public void KappaTestConstructorTest2()
{
int[,] matrix =
{
{ 317, 23, 0, 0 },
{ 61, 120, 0, 0 },
{ 2, 4, 60, 0 },
{ 35, 29, 0, 8 },
};
GeneralConfusionMatrix a = new GeneralConfusionMatrix(matrix);
Assert.AreEqual(0.7663, a.OverralAgreement, 1e-4);
Assert.IsFalse(double.IsNaN(a.OverralAgreement));
Assert.AreEqual(0.4087, a.ChanceAgreement, 1e-5);
Assert.IsFalse(double.IsNaN(a.ChanceAgreement));
KappaTest target = new KappaTest(a);
Assert.AreEqual(0.605, target.Kappa, 1e-3);
Assert.AreEqual(a.Kappa, target.Kappa);
Assert.IsFalse(double.IsNaN(target.Kappa));
Assert.AreEqual(0.00073735, target.Variance, 1e-5);
Assert.IsFalse(double.IsNaN(target.Variance));
Assert.AreEqual(22.272, target.Statistic, 0.15);
Assert.IsFalse(double.IsNaN(target.Statistic));
Assert.AreEqual(0.658, target.Confidence.Max, 1e-3);
Assert.AreEqual(0.552, target.Confidence.Min, 1e-3);
Assert.AreEqual(0.0, target.PValue, 1e-6);
Assert.IsFalse(double.IsNaN(target.PValue));
Assert.IsTrue(target.Significant);
}
public void KappaTestConstructorTest()
{
int[,] matrix =
{
{ 44, 5, 1 },
{ 7, 20, 3 },
{ 9, 5, 6 },
};
GeneralConfusionMatrix a = new GeneralConfusionMatrix(matrix);
Assert.AreEqual(0.4915, a.Kappa, 1e-4);
Assert.IsFalse(double.IsNaN(a.Kappa));
KappaTest target = new KappaTest(a);
Assert.AreEqual(0.0777, target.StandardError, 1e-2);
Assert.IsFalse(double.IsNaN(target.StandardError));
Assert.AreEqual(0.3393, target.Confidence.Min, 1e-2);
Assert.AreEqual(0.6437, target.Confidence.Max, 1e-2);
}
public void KappaTestConstructorTest1()
{
// Example from Ientilucci, Emmett (2006). "On Using and Computing the Kappa Statistic".
// Available on: http://www.cis.rit.edu/~ejipci/Reports/On_Using_and_Computing_the_Kappa_Statistic.pdf
// This paper uses the delta method approximation
// for computing the kappa variance.
int[,] matrix1 =
{
{ 317, 23, 0, 0 },
{ 61, 120, 0, 0 },
{ 2, 4, 60, 0 },
{ 35, 29, 0, 8 },
};
int[,] matrix2 =
{
{ 377, 79, 0, 0 },
{ 2, 72, 0, 0 },
{ 33, 5, 60, 0 },
{ 3, 20, 0, 8 },
};
GeneralConfusionMatrix a = new GeneralConfusionMatrix(matrix1);
GeneralConfusionMatrix b = new GeneralConfusionMatrix(matrix2);
Assert.AreEqual(0.7663, a.OverallAgreement, 1e-4);
Assert.AreEqual(0.7845, b.OverallAgreement, 1e-4);
Assert.AreEqual(0.4087, a.ChanceAgreement, 1e-4);
Assert.AreEqual(0.47986, b.ChanceAgreement, 1e-4);
double kA = a.Kappa;
double kB = b.Kappa;
double varA = KappaTest.DeltaMethodKappaVariance(a);
double varB = KappaTest.DeltaMethodKappaVariance(b);
TwoMatrixKappaTest target = new TwoMatrixKappaTest(kA, varA, kB, varB);
Assert.AreEqual(TwoSampleHypothesis.ValuesAreDifferent, target.Hypothesis);
Assert.AreEqual(DistributionTail.TwoTail, target.Tail);
// Compare Kappas
Assert.AreEqual(0.605, target.EstimatedValue1, 1e-3);
Assert.IsFalse(double.IsNaN(a.Kappa));
Assert.AreEqual(0.586, target.EstimatedValue2, 1e-3);
Assert.IsFalse(double.IsNaN(b.Kappa));
// Compare variances:
Assert.AreEqual(0.00073735, target.Variance1, 1e-7);
Assert.IsFalse(double.IsNaN(a.Variance));
Assert.AreEqual(0.00087457, target.Variance2, 1e-7);
Assert.IsFalse(double.IsNaN(b.Variance));
Assert.AreEqual(0.475, target.Statistic, 1e-3);
Assert.IsFalse(double.IsNaN(target.Statistic));
Assert.IsFalse(target.Significant);
}
public void KappaTestConstructorTest2()
{
// Example from Congalton
int[,] matrix1 = // pg 108
{
{ 65, 4, 22, 24 },
{ 6, 81, 5, 8 },
{ 0, 11, 85, 19 },
{ 4, 7, 3, 90 },
};
GeneralConfusionMatrix a = new GeneralConfusionMatrix(matrix1);
Assert.AreEqual(115, a.RowTotals[0]);
Assert.AreEqual(100, a.RowTotals[1]);
Assert.AreEqual(115, a.RowTotals[2]);
Assert.AreEqual(104, a.RowTotals[3]);
Assert.AreEqual(75, a.ColumnTotals[0]);
Assert.AreEqual(103, a.ColumnTotals[1]);
Assert.AreEqual(115, a.ColumnTotals[2]);
Assert.AreEqual(141, a.ColumnTotals[3]);
int[,] matrix2 = // pg 109
{
{ 45, 4, 12, 24 },
{ 6, 91, 5, 8 },
{ 0, 8, 55, 9 },
{ 4, 7, 3, 55 },
};
GeneralConfusionMatrix b = new GeneralConfusionMatrix(matrix2);
Assert.AreEqual(85, b.RowTotals[0]);
Assert.AreEqual(110, b.RowTotals[1]);
Assert.AreEqual(72, b.RowTotals[2]);
Assert.AreEqual(69, b.RowTotals[3]);
Assert.AreEqual(55, b.ColumnTotals[0]);
Assert.AreEqual(110, b.ColumnTotals[1]);
Assert.AreEqual(75, b.ColumnTotals[2]);
Assert.AreEqual(96, b.ColumnTotals[3]);
// Check overall accuracy
Assert.AreEqual(0.74, a.OverallAgreement, 0.005);
Assert.AreEqual(0.73, b.OverallAgreement, 0.005);
double kA = a.Kappa;
double kB = b.Kappa;
double varA = KappaTest.DeltaMethodKappaVariance(a);
double varB = KappaTest.DeltaMethodKappaVariance(b);
// Create the test
TwoMatrixKappaTest target = new TwoMatrixKappaTest(kA, varA, kB, varB);
Assert.AreEqual(TwoSampleHypothesis.ValuesAreDifferent, target.Hypothesis);
Assert.AreEqual(DistributionTail.TwoTail, target.Tail);
// Compare Kappas (pg 109)
Assert.AreEqual(0.65, target.EstimatedValue1, 0.05);
Assert.IsFalse(double.IsNaN(a.Kappa));
Assert.AreEqual(0.64, target.EstimatedValue2, 0.05);
Assert.IsFalse(double.IsNaN(b.Kappa));
// Compare variances:
Assert.AreEqual(0.0007778, target.Variance1, 1e-7);
Assert.IsFalse(double.IsNaN(a.Variance));
Assert.AreEqual(0.0010233, target.Variance2, 1e-7);
Assert.IsFalse(double.IsNaN(b.Variance));
Assert.AreEqual(0.3087, target.Statistic, 1e-5);
Assert.IsFalse(double.IsNaN(target.Statistic));
Assert.IsFalse(target.Significant);
}
/// <summary>
/// 模拟
/// </summary>
public void Simulate()
{
this.ChartCountOfTypes = this.landInfo.NumOfLandUseTypes;
this.ChartCellCountArr = new int[this.ChartCountOfTypes];
this.InitialChart();
this.BeginCityCnt = this.GetCityCnt();
int nowCityCnt = this.BeginCityCnt;
// 计算lg值
this.lgBuffer = GetLgBuffer();
int times = 0;
int numOfAll = 0;
while (times < this.timeOfSimulate)
{
if (nowCityCnt > this.EndCityCnt)
{
break;
}
// 清空图表存储数组
this.ClearChartCellCountArr();
times++;
int numOfChange = 0;
#region parallel for
Parallel.For(0, height * width, pos =>
{
int row = pos / width;
int col = pos % width;
middleBuffer[pos] = beginBuffer[pos];
if (this.landInfo.IsExistInConvertableInfos(this.beginBuffer[pos]))
{
double globalValue = this.lgBuffer[pos];
double localValue = GetNeighbourAffect(row, col, this.sizeOfNeighbour);
double randomValue = Math.Pow(-Math.Log(ThreadLocalRandom.NextDouble()), this.alpha);
if (globalValue * localValue * randomValue > threshold)
{
this.middleBuffer[pos] = this.landInfo.UrbanInfos[0].LandUseTypeValue;
numOfChange++;
numOfAll++;
nowCityCnt++;
}
// 修改图表计数数组
}
AddChartCellCountArr(middleBuffer[pos]);
});
#endregion
#region sequential
// for (int row = 0; row < height; row++)
//{
// for (int col = 0; col < width; col++)
// {
// int pos = row * width + col;
// this.middleBuffer[pos] = this.beginBuffer[pos];
// if (this.landInfo.IsExistInConvertableInfos(this.beginBuffer[pos]))
// {
// double globalValue = this.lgBuffer[pos];
// double localValue = GetNeighbourAffect(row, col, this.sizeOfNeighbour);
// double randomValue = Math.Pow(-Math.Log(rnd.NextDouble()), this.alpha);
// if (globalValue * localValue * randomValue > threshold)
// {
// this.middleBuffer[pos] = this.landInfo.UrbanInfos[0].LandUseTypeValue;
// numOfChange++;
// numOfAll++;
// nowCityCnt++;
// }
// // 修改图表计数数组
// }
// AddChartCellCountArr(middleBuffer[pos]);
// }
//}
#endregion
updateConsoleEvent("-------");
updateConsoleEvent("当前城市数目： " + nowCityCnt);
updateConsoleEvent("这次增长了" + numOfChange + "个元胞");
updateConsoleEvent("-------");
this.updateImageEvent(this.middleBuffer, this.width, this.height, this.landInfo, times);
this.updateChartEvent(this.ChartCellCountArr, times++, this.landInfo);
// 将middle buffer的值复制会begin buffer
for (int pos = 0; pos < width * height; pos++)
{
this.beginBuffer[pos] = this.middleBuffer[pos];
}
}
this.updateConsoleEvent("模拟结束");
this.updateConsoleEvent("总共增长了" + numOfAll + "个元胞");
var kappa = new KappaTest(this.endBuffer, this.beginBuffer, width, height, landInfo);
var kappaVal = kappa.GetVal();
this.updateConsoleEvent("kappa值 : " + kappaVal);
this.simulateEndEvent(this);
}
public void Simulate(object obj)
{
int nowCityCnt = this.BeginCityCnt;
int times = (int)obj;
// 开始模拟
int cnt = 0;
double[] middleBuffer = new double[width * height];
while (cnt < times)
{
int addedCity = 0;
if (nowCityCnt > this.EndCityCnt)
{
break;
}
// 清空图表计数数组
this.ClearChartCellCountArr();
#region parallel for
Parallel.For(0, height * width, pos =>
{
int row = pos / width;
int col = pos % width;
middleBuffer[pos] = beginBuffer[pos];
double type = beginBuffer[pos];
// 只考虑可以转化为城市的栅格
if (this.landInfo.IsExistInConvertableInfos(type))
{
List <double> input = (from buffer in driveBufferList
select buffer[pos]).ToList <double>();
input.Add(GetNeighbourAffect(beginBuffer, width, height, row, col, 3));
double[] inputArr = input.ToArray <double>();
int newType = func(inputArr);
// 如果转化为城市
if (newType == 1 && ThreadLocalRandom.NextDouble() < 0.1)
{
middleBuffer[pos] = this.landInfo.UrbanInfos[0].LandUseTypeValue;
addedCity++;
nowCityCnt++;
}
}
// 修改图表计数数组
AddChartCellCountArr(middleBuffer[pos]);
});
#endregion
#region sequential for
//for (int row = 0; row < height; row++)
//{
// for (int col = 0; col < width; col++)
// {
// int pos = row * width + col;
// middleBuffer[pos] = beginBuffer[pos];
// double type = beginBuffer[pos];
// // 只考虑可以转化为城市的栅格
// if (this.landInfo.IsExistInConvertableInfos(type))
// {
// List<double> input = (from buffer in driveBufferList
// select buffer[pos]).ToList<double>();
// input.Add(GetNeighbourAffect(beginBuffer, width, height, row, col, 3));
// double[] inputArr = input.ToArray<double>();
// int newType = func(inputArr);
// // 如果转化为城市
// if (newType == 1 && rnd.NextDouble() < 0.1)
// {
// middleBuffer[pos] = this.landInfo.UrbanInfos[0].LandUseTypeValue;
// addedCity++;
// nowCityCnt++;
// }
// }
// // 修改图表计数数组
// AddChartCellCountArr(middleBuffer[pos]);
// }
//}
#endregion
// 将结果复制会outputBuffer
for (int pos = 0; pos < width * height; pos++)
{
beginBuffer[pos] = middleBuffer[pos];
}
updateConsoleEvent("-------");
updateConsoleEvent("当前城市数目:" + nowCityCnt);
updateChartEvent(ChartCellCountArr, cnt + 1, landInfo);
updateImageEvent(beginBuffer, width, height, this.landInfo, cnt); // 通知主线程绘图
updateConsoleEvent("新增： " + addedCity); // 通知主线程输出控制台消息
cnt++;
}
updateConsoleEvent("模拟结束");
var kappa = new KappaTest(this.endBuffer, this.beginBuffer, width, height, landInfo);
var kappaVal = kappa.GetVal();
this.updateConsoleEvent("kappa值 : " + kappaVal);
}
/// <summary>
/// 模拟
/// </summary>
public void Simulate(int times)
{
updateConsoleEvent("--------- 模拟开始");
updateConsoleEvent("初始城市数目:" + this.BeginCityCnt);
updateConsoleEvent("目标城市数目：" + this.EndCityCnt);
updateConsoleEvent("---------");
Random rnd = new Random();
int cnt = 0;
int numOfChangesOneTime = 0;
double[] middleBuffer = new double[width * height];
int nowCityCnt = this.BeginCityCnt;
while (cnt < times)
{
if (nowCityCnt > this.EndCityCnt)
{
break;
}
// 清空图表存储数组
this.ClearChartCellCountArr();
#region parallel for
Parallel.For(0, height * width, pos =>
{
int row = pos / width;
int col = pos % width;
middleBuffer[pos] = beginBuffer[pos];
// null value 跳过
if (beginBuffer[pos] < 0 || Math.Abs(beginBuffer[pos] - this.landInfo.NullInfo.LandUseTypeValue) < Double.Epsilon)
{
return;
}
var input = getOneInput(new Cell()
{
row = row, col = col
});
double[] answer = this.network.Compute(input);
int actual;
double max = answer.Max(out actual);
//double ra = (double)(1 + Math.Pow(-Math.Log(ThreadLocalRandom.NextDouble(), Math.E), this.alpha)); // -------todo 这个随机数太大了？？
double ra = ThreadLocalRandom.NextDouble();
if (max * ra > this.threshold)
{
// 转换控制
int srcIdx = 0;
for (int i = 0; i < this.landInfo.NumOfLandUseTypes; i++)
{
if (middleBuffer[pos] == this.landInfo.AllTypes[i].LandUseTypeValue)
{
srcIdx = i;
}
}
if (this.TransformControlMatrix[srcIdx, actual] == 0)
{
// 不转换
AddChartCellCountArr(middleBuffer[pos]);
return;
}
middleBuffer[pos] = landInfo.AllTypes[actual].LandUseTypeValue;
if (landInfo.AllTypes[actual].LandUseTypeValue != beginBuffer[pos])
{
if (actual == this.landInfo.UrbanIndex)
{
nowCityCnt++;
}
numOfChangesOneTime++;
}
}
// 修改图表计数数组
AddChartCellCountArr(middleBuffer[pos]);
});
#endregion
#region sequential for
//for (int row = 0; row < height; row++)
//{
// for (int col = 0; col < width; col++)
// {
// int pos = row * width + col;
// middleBuffer[pos] = beginBuffer[pos];
// // null value 跳过
// if (beginBuffer[pos] < 0 || Math.Abs(beginBuffer[pos] - this.landInfo.NullInfo.LandUseTypeValue) <Double.Epsilon)
// {
// continue;
// }
// var input = getOneInput(new Cell() { row = row, col = col });
// double[] answer = this.network.Compute(input);
// int actual;
// double max = answer.Max(out actual);
// double ra = (double)(1 + Math.Pow(-Math.Log(rnd.NextDouble(), Math.E), this.alpha)); -------todo 这个随机数太大了？？
// //double ra = rnd.NextDouble();
// if (max * ra > this.threshold)
// {
// // 转换控制
// int srcIdx = 0;
// for (int i = 0; i < this.landInfo.NumOfLandUseTypes; i++)
// {
// if (middleBuffer[pos] == this.landInfo.AllTypes[i].LandUseTypeValue)
// {
// srcIdx = i;
// }
// }
// if (this.TransformControlMatrix[srcIdx, actual] == 0)
// {
// // 不转换
// AddChartCellCountArr(middleBuffer[pos]);
// continue;
// }
// middleBuffer[pos] = landInfo.AllTypes[actual].LandUseTypeValue;
// if (landInfo.AllTypes[actual].LandUseTypeValue != beginBuffer[pos])
// {
// if (actual == this.landInfo.UrbanIndex)
// {
// nowCityCnt++;
// }
// numOfChangesOneTime++;
// }
// }
// // 修改图表计数数组
// AddChartCellCountArr(middleBuffer[pos]);
// }
//}
#endregion
this.beginBuffer = middleBuffer;
updateConsoleEvent("------");
updateConsoleEvent("当前城市数目:" + nowCityCnt);
updateImageEvent(this.beginBuffer, width, height, this.landInfo, cnt); // draw
updateChartEvent(this.ChartCellCountArr, cnt + 1, landInfo);
middleBuffer = new double[width * height];
updateConsoleEvent("转化了:" + numOfChangesOneTime);
updateConsoleEvent("------");
numOfChangesOneTime = 0;
cnt++;
}
this.updateConsoleEvent("---------------模拟结束------------");
var kappa = new KappaTest(this.endBuffer, this.beginBuffer, width, height, landInfo);
var kappaVal = kappa.GetVal();
this.updateConsoleEvent("kappa值 : " + kappaVal);
}
/// <summary>
/// 模拟
/// </summary>
public void Simulate(int times)
{
int nowCityCnt = this.nowCityCount;
int cnt = 0;
int numOfChangesOneTime = 0;
double[] middleBuffer = new double[width * height];
while (cnt < times)
{
if (this.nowCityCount > this.targetCityCnt)
{
break;
}
// 清空图表存储数组
this.ClearChartCellCountArr();
#region parallel for
Parallel.For(0, height * width, pos =>
{
int row = pos / width;
int col = pos % width;
middleBuffer[pos] = beginBuffer[pos];
if (!IsValid(pos)) // 无效值 跳过
{
return;
}
if (middleBuffer[pos] == this.landInfo.WaterInfo.LandUseTypeValue) // 水体跳过
{
AddChartCellCountArr(middleBuffer[pos]);
return;
}
if (this.landInfo.IsExistInUrbanInfos(middleBuffer[pos])) // 城市跳过
{
AddChartCellCountArr(middleBuffer[pos]);
return;
}
var input = getOneInput(new Cell()
{
row = row, col = col
});
double[] propabilities = new double[this.landInfo.AllTypes.Count];
alglib.dfprocess(this.Forest, input, ref propabilities);
for (int i = 0; i < propabilities.Length; i++)
{
double ra = (double)(1 + Math.Pow(-Math.Log(ThreadLocalRandom.NextDouble(), Math.E), this.alpha));
double neighbour = GetNeighbourAffect(row, col, this.landInfo.AllTypes[i].LandUseTypeValue, this.sizeOfNeighbour);
propabilities[i] = propabilities[i] * ra * neighbour;
}
int idx = getIdxFromPropArr(propabilities);
if (idx == this.landInfo.UrbanIndex)
{
// Interlocked.Add(ref nowCityCnt, 1);
// this.nowCityCount++;
nowCityCnt++;
}
if (this.landInfo.AllTypes[idx].LandUseTypeValue == this.landInfo.WaterInfo.LandUseTypeValue) // target 不能为水体
{
AddChartCellCountArr(middleBuffer[pos]);
return;
}
// 转换控制
int srcIdx = 0;
for (int i = 0; i < this.landInfo.NumOfLandUseTypes; i++)
{
if (middleBuffer[pos] == this.landInfo.AllTypes[i].LandUseTypeValue)
{
srcIdx = i;
}
}
if (this.transformControlMatrix[srcIdx, idx] == 0)
{
// 不转换
AddChartCellCountArr(middleBuffer[pos]);
return;
}
middleBuffer[pos] = this.landInfo.AllTypes[idx].LandUseTypeValue;
if (middleBuffer[pos] != this.beginBuffer[pos])
{
numOfChangesOneTime++;
}
// 修改图表计数数组
AddChartCellCountArr(middleBuffer[pos]);
});
#endregion
this.nowCityCount = nowCityCnt;
this.updateConsoleEvent("---当前城市数目: " + this.nowCityCount);
this.beginBuffer = middleBuffer;
if (updateImageEvent != null)
{
this.updateImageEvent(this.beginBuffer, this.width, this.height, this.landInfo, cnt);
}
this.updateChartEvent(this.ChartCellCountArr, cnt + 1, landInfo);
middleBuffer = new double[width * height];
if (updateConsoleEvent != null)
{
this.updateConsoleEvent("转化了:" + numOfChangesOneTime);
}
numOfChangesOneTime = 0;
cnt++;
}
if (updateConsoleEvent != null)
{
this.updateConsoleEvent("模拟结束");
}
var kappa = new KappaTest(this.endBuffer, this.beginBuffer, width, height, landInfo);
var kappaVal = kappa.GetVal();
this.updateConsoleEvent("kappa值 : " + kappaVal);
}