You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Based on looking at the code pdcDist works for multivariate time series if X is a 3d array but the documentation only refers to the univariate case where X is a matrix. Suggest making documentation consistent with code and providing an example of its use with multivariate time series. Also couldn't find how the distance between two univariate series is extended to handle multivariate.
Also the jstatsoft paper refers to using the likelihood ratio to determine the number of clusters. If this is implemented suggest documenting how to get the number of clusters using the package. If t his has not been implemented it would be a useful addition.
The text was updated successfully, but these errors were encountered:
Thanks, @ggrothendieck for pointing this out. It is correct that the documentation for using 3d arrays is only in the Rd file for pdclust but not in pdcDist. Here is a new vignette that demonstrates usage of multivariate time series: https://brandmaier.github.io/pdc/articles/Multivariate.html
The hypothesis test is implemented in `hypothesis.test.R" but not exposed to public because it lacks a proper interface. I will keep this issue open as a reminder to implement this.
Based on looking at the code pdcDist works for multivariate time series if X is a 3d array but the documentation only refers to the univariate case where X is a matrix. Suggest making documentation consistent with code and providing an example of its use with multivariate time series. Also couldn't find how the distance between two univariate series is extended to handle multivariate.
Also the jstatsoft paper refers to using the likelihood ratio to determine the number of clusters. If this is implemented suggest documenting how to get the number of clusters using the package. If t his has not been implemented it would be a useful addition.
The text was updated successfully, but these errors were encountered: