Closed cdrnet closed 3 years ago
Bump. Any plans @cdrnet? @mchandschuh
How would an API design for this ideally look like, that would support your use case in a clean way?
Ideally the API would support usage similar to this python implementation.
Following along the lines of the Correlation
class, I could see something like the following in C#:
var result = Cointegration.Johansen(Matrix<double> matrix, int order, int laggedDifferences);
This suggested API makes sense.
FWIW, the suggested api is reasonably consistent with the R package for Johansen Cointegration Test (urca). It is a little bit different than the matlab econometrics toolbox version of Johansen Cointegration, but that is probably ok because the matlab api is older and offers a lot more complex control over the implementation details which likely wouldn't be built into a first version of JCI in mathnet.
On Fri, Sep 18, 2015 at 2:30 AM, Michael notifications@github.com wrote:
Ideally the API would support usage similar to this python https://github.com/josef-pkt/statsmodels/blob/coint/statsmodels/tsa/johansen.py implementation.
Following along the lines of the Correlation class, I could see something like the following in C#:
var result = Cointegration.Johansen(Matrix
matrix, int order, int laggedDifferences); — Reply to this email directly or view it on GitHub https://github.com/mathnet/mathnet-numerics/issues/141#issuecomment-141356712 .
Any updates on this? It would be really cool if any Cointegration support was added. Johansen or Augmented Dickey Fuller (ADF)!
From: https://mathnetnumerics.codeplex.com/workitem/5730