Re: [R] TSclust multivariate time series clustering

Hello, Although there is no R package available (we did not think of it), if you want a Gaussian-mixture=model-based approach, you may look at the paper: "Model-Based Clustering of Regression Time Series Data via APECM—An AECM Algorithm Sung to an Even Faster Beat" by Wei-Chen Chen and Ranjan

Re: [R] TSclust multivariate time series clustering

You can also try kml3d. You can either use some default distances or define your own. http://www.jstatsoft.org/v65/i04/paper Christophe -- View this message in context: http://r.789695.n4.nabble.com/TSclust-multivariate-time-series-clustering-tp4711517p4711555.html Sent from the R help mailin

Re: [R] TSclust multivariate time series clustering

I suggest you look at this paper: http://www.jstatsoft.org/v62/i01/paper -Roy > On Aug 26, 2015, at 1:02 AM, GloriaSolano > wrote: > > Hello, > > I am trying to cluster multivariate time series with the R package TSclust. > I have a dataset of 45 companies with 10 years information on 6 vari

[R] TSclust multivariate time series clustering

Hello, I am trying to cluster multivariate time series with the R package TSclust. I have a dataset of 45 companies with 10 years information on 6 variables. I understand from the TSclust package manual that the first step is choosing the dissimilarity meassure we are going to use. However I canno