Using MathNet.Numerics to fit regression line through origin

Question

I have two double arrays (double[] XValues, YValues) which contain physiological data.

On these data I currently perform an 'unconstrained' linear regression using:

Tuple<double, double> r = Fit.Line(XValues, YValues);
double YIntercept = r.Item1;
double Slope = r.Item2;

There are sound (physiological) reasons for allowing the user to examine the line of best fit when regression through the origin is enforced.

How can this be done using MathNet?

Answer 1

除了Fit.Line之外，还有一个更通用的Fit.LinearCombination函数。由于前者基本上试图在p0中找到p1和y : x -> p0 * 1.0 + p1 * x，因此它也可以写为Fit.LinearCombination(XValues, YValues, x => 1.0, x => x)。

您正在寻找原点的一条线，其中p0被强制为零。因此，我们的等式简化为y : x -> p1 * x，可以使用Fit.LinearCombination(XValues, YValues, x => x)计算。