All the routines are supplied as is. The routines almost always rely on a matrix class with the basic linear algebra operations and standard utility methods defined. Exactly what is needed should be clear from the code. Also, in some cases optimization routines can be required. All of this can be found in standard references such as Press et al (2002). A translation to, and implementation in, C# is straightforward.
Implementation of linear regression model. Some features in the code are:
HC/HAC standard errors
Heteroskedasticity consistent variance/covaraince matrix estimation, Heteroskedasticity and Autocorrelation consistent variance/covaraince matrix estimation with Bartlett, Parzen, Quadrtic Spectral and Truncated kernel.
Specification and Diagnostic tests
Standard Wald test, RESET test, Omitted-variable test, Residual serial correlation test, Breusch-Pagan test, White's test, ARCH test.
The code is in available here: LinReg.cs
Kalman filter and smoother applied to state space model. File: StateSpace.cs
In modelling time series, it is often improtant to know if a series contains a unit root. To find out, a range of tests can be used. Three different unit root tests are implemented in the C# code supplied: The Augmented Dickey-Fuller test, the Phillips-Perron test, an the KPSS test. File: URTest.cs
It is seldom the case that all series that are of interest for modelling have the same frequency. MIDAS models can handle this. Two different MIDAS specifications (step and Almon) are available in the code. File: MIDAS.cs
In many settings it is of interest to estimate the scalled long-run variance. Different kernels can be used. The code implements the following kernels: Truncated, Bartlett, Quadratic Spectral, and Parzen. File: LRVar.cs
Press, William H., Teukolsky, Saul A., Vetterling, William T. and Flannery, Brian P. (2002) Numerical Recipes in C++, 2nd edition, Cambridge University Press.