Anderson Darling Test Constructor
Definition
Assembly: Extreme.Numerics (in Extreme.Numerics.dll) Version: 8.1.23
Overload List
Anderson | Constructs a new Anderson-Darling test of normality. |
Anderson | Constructs a new Anderson-Darling test of normality. |
Anderson | Constructs a new Anderson-Darling test for the specified distribution. |
Anderson | Constructs a new Anderson-Darling test for the specified distribution function. |
Anderson | Constructs a new Anderson-Darling test of normality. |
AndersonDarlingTest
public AndersonDarlingTest()
AndersonDarlingTest(Vector<Double>)
Remarks
The Anderson-Darling test works on any sample. The sample does not have to be normalized, for example.
AndersonDarlingTest(Vector<Double>, ContinuousDistribution)
public AndersonDarlingTest(
Vector<double> sample,
ContinuousDistribution distributionType
)
Parameters
- sample Vector<Double>
- A vector.
- distributionType ContinuousDistribution
- The Type of the reference distribution.
Remarks
The Anderson-Darling test works on any sample. The sample does not have to be normalized.
Some distributions have specialized equations for computing p-values: NormalDistribution, LognormalDistribution, ExponentialDistribution, GumbelDistribution (largest value only), LogisticDistribution.
Note that for these types of reference distributions, the test will return results that are different from an Anderson-Darling test created using the distribution's cumulative distribution function (CDF) directly. This is because the algorithm used to compute the p-values for the Anderson-Darling distribution is different.
For other distributions, the cumulative distribution function is fitted directly.
AndersonDarlingTest(Vector<Double>, Func<Double, Double>)
public AndersonDarlingTest(
Vector<double> sample,
Func<double, double> cdf
)
Parameters
- sample Vector<Double>
- A vector.
- cdf Func<Double, Double>
- A function that evaluates the cumulative distribution function (CD) of the reference distribution.
Remarks
This algorithm is based on a method by Marsaglia for computing the distribution function of the Anderson-Darling statistic for a uniform distribution.
Note that for some types of reference distributions, this method will return results that are different from an Anderson-Darling test created using the specific distribution object instead of its distribution function.
AndersonDarlingTest(Vector<Double>, Double, Double)
public AndersonDarlingTest(
Vector<double> sample,
double mean,
double standardDeviation
)
Parameters
- sample Vector<Double>
- A vector.
- mean Double
- The mean of the normal distribution to test against.
- standardDeviation Double
- The standard deviation of the normal distribution to test against.
Remarks
Use this constructor to create an AndersonDarlingTest object that can be used to test whether a sample follows a specific normal distribution.
Exceptions
ArgumentOutOfRangeException | standardDeviation is less than or equal to zero. |