| public static LinearRegression ( double xVals, double yVals, int inclusiveStart, int exclusiveEnd, double &rsquared, double &yintercept, double &slope ) : void | ||
| xVals | double | The x-axis values. |
| yVals | double | The y-axis values. |
| inclusiveStart | int | The inclusive inclusiveStart index. |
| exclusiveEnd | int | The exclusive exclusiveEnd index. |
| rsquared | double | The r^2 value of the line. |
| yintercept | double | The y-intercept value of the line (i.e. y = ax + b, yintercept is b). |
| slope | double | The slop of the line (i.e. y = ax + b, slope is a). |
| return | void | |
public void addSampleToRegression(int sample)
{
bufferSamples.Add(sample);
DateTime now = DateTime.UtcNow;
int nowInMillis = (((now.Hour * 60) + now.Minute) * 60 + now.Second) * 1000 + now.Millisecond;
times.Add(nowInMillis);
if (bufferSamples.Count > max_num_of_samples)
{
bufferSamples.RemoveAt(0);
times.RemoveAt(0);
}
Regression.LinearRegression(times.ToArray(), bufferSamples.ToArray().Select(i => (double)i).ToArray(), 0, times.Count, out rsquared, out yintercept, out slope);
}
private void PerformLinear_Click(object sender, RoutedEventArgs e)
{
double[,] points = getPointSeries();
if (points == null)
{
MessageBox.Show("No points :(");
return;
}
Coefficients aAndb = Regression.LinearRegression(points);
PlotModel m = new PlotModel {
};
m.Axes.Add(new LinearAxis {
Position = AxisPosition.Bottom
});
m.Axes.Add(new LinearAxis {
Position = AxisPosition.Left, Maximum = 10, Minimum = -10
});
m.Series.Add(new FunctionSeries(x => aAndb.a * x + aAndb.b, -10f, 10f, 0.1f, null));
LineSeries series1 = new LineSeries
{
LineStyle = LineStyle.None,
MarkerType = MarkerType.Circle,
MarkerSize = 4,
MarkerStroke = OxyColors.White
};
for (int i = points.GetLength(0) - 1; i >= 0; i--)
{
series1.Points.Add(new DataPoint(points[i, 0], points[i, 1]));
}
m.Series.Add(series1);
Graph.Model = m;
}
private void PlotBreakEvenPoint()
{
plot4.Clear();
Finances finances = GameController.instance.player.finances;
List <int> priceData = new List <int>();
priceData.AddRange(finances.revenue);
priceData.Sort();
List <int> quantityData = finances.GetQuantityData();
Regression regression = new Regression();
float a = 0; float b = 0; float rSq = 0;
//regression.LinearRegression(quantityData.ToArray(), finances.revenue.ToArray(),out a, out b, out rSq);
List <float> regressionData = new List <float>();
float xSize = (quantityData.Max() - quantityData.Min()) / quantityData.Count;
List <float> quantityNormalizedData = new List <float>()
{
0
};
for (int i = 0; i < quantityData.Count; i++)
{
quantityNormalizedData.Add(quantityNormalizedData.LastOrDefault() + xSize);
}
float ySize = (priceData.Max() - priceData.Min()) / priceData.Count;
List <float> regressionDataNormalized = new List <float>()
{
0
};
for (int i = 0; i < priceData.Count; i++)
{
regressionDataNormalized.Add(regressionDataNormalized.LastOrDefault() + ySize);
}
regressionDataNormalized.Remove(regressionDataNormalized.LastOrDefault());
// regressionData.AddRange(regression.Function(quantityNormalizedData, a, b));
List <int> x = new List <int>()
{
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12
};
LineGraphVisual originalVisual = new LineGraphVisual(containers[3], dotSprite, 12, false, new Color(1, 1, 1, 0.0f), Color.white);
LineGraphVisual revenueVisual = new LineGraphVisual(containers[3], dotSprite, 12, false, Color.cyan, Color.white);
LineGraphVisual regressionVisual = new LineGraphVisual(containers[3], dotSprite, 12, false, Color.yellow, Color.white);
LineGraphVisual expencesRegressionVisual = new LineGraphVisual(containers[3], dotSprite, 12, false, Color.red, Color.red);
List <float> extrems = new List <float>();
List <float> expencesRegression = new List <float>();
if (finances.revenue.Max() >= finances.staticExpences.Max())
{
extrems = plot4.ShowPlot(finances.revenue, originalVisual, 14, containers[3], 0, 0, null, displayGrid: false);
}
else
{
extrems = plot4.ShowPlot(finances.staticExpences, originalVisual, -1, containers[3], 0, 0, displayGrid: false);
}
plot4.ShowPlot(regressionDataNormalized, revenueVisual, 14, containers[3], extrems[0], extrems[1],
(int i) => Mathf.RoundToInt(quantityNormalizedData[i]).ToString(), (float f) => String.Format("{0:n0} $", Mathf.RoundToInt(f)), displayGrid: true);
List <int> yearExpences = finances.GetOverYearExpences();
if (yearExpences.Count < x.Count)
{
for (int i = yearExpences.Count; i < x.Count; i++)
{
yearExpences.Add(yearExpences.LastOrDefault());
}
}
regression.LinearRegression(x.ToArray(), yearExpences.ToArray(), out a, out b, out rSq);
// take last 12 elements and delete the last one in order to make regression. Adjust x accordingly
expencesRegression = regression.Function(x.ConvertAll(p => (float)p), a, b);
Debug.Log("-----Linear Regression-----");
Debug.Log("A: " + a + " B: " + b + "Rsquare: " + rSq);
plot4.ShowPlot(expencesRegression, expencesRegressionVisual, -1, containers[3], extrems[0], extrems[1], displayGrid: false);
}
private bool FitModel(Steema.TeeChart.Styles.Series source, ref double[] yhat, out double[] coeffs, int modelindex)
{
double[] y = new double[source.Count];
double[] x = new double[source.Count];
double[] w = new double[source.Count];
bool validmodel = true;
switch (modelindex)
{
case 0:
// y(x)=a*Exp(b*x)
// linear model: ln(y) = b*x + ln(a)
for (int i = 0; i < source.Count; i++)
{
x[i] = source.notMandatory[i];
y[i] = Math.Log(source.mandatory[i]);
w[i] = source.mandatory[i] * source.mandatory[i];
}
Regression.LinearRegression(source.Count, x, y, w, out coeffs);
coeffs[0] = Math.Exp(coeffs[0]);
for (int i = 0; i < source.Count; i++)
{
yhat[i] = coeffs[0] * Math.Exp(coeffs[1] * x[i]);
}
break;
case 1:
// Power model y=a*x^b
// linear model : ln(y) = b*ln(x) + ln(a)
for (int i = 0; i < source.Count; i++)
{
x[i] = Math.Log(source.notMandatory[i]);
y[i] = Math.Log(source.mandatory[i]);
w[i] = source.mandatory[i] * source.mandatory[i];
}
Regression.LinearRegression(source.Count, x, y, w, out coeffs);
coeffs[0] = Math.Exp(coeffs[0]);
for (int i = 0; i < source.Count; i++)
{
yhat[i] = coeffs[0] * Math.Pow(source.notMandatory[i], coeffs[1]);
}
break;
case 2:
// Logarithmic model y = b*ln(x)+a
for (int i = 0; i < source.Count; i++)
{
x[i] = Math.Log(source.notMandatory[i]);
y[i] = source.mandatory[i];
w[i] = source.mandatory[i] * source.mandatory[i];
}
Regression.LinearRegression(source.Count, x, y, w, out coeffs);
for (int i = 0; i < source.Count; i++)
{
yhat[i] = coeffs[1] * Math.Log(source.notMandatory[i]) + coeffs[0];
}
break;
default: coeffs = new double[0]; break;
}
return(validmodel);
}