My data set are like this

DF1     DF2      DF3     DF4  ....... DF50
 A       B        B       A            C
 A       A        C       B            B
 C       A        A       B            B
 .       .        .       .            .
 .       .        .       .            .
 .       .        .       .            .
 F       G        E       A            G

Here are five 5 states (A-G) for each of the DF. It is not mandatory to have all the letter in each of the DF.

I want to do sequence analysis (sequence clustering). I have already done a method on it. By using 1st order Markov chain I have created transition probability for each of them. Then I have applied Frobenius norm on that transition probability to calculate distance. By using cutoff value(trial and error) I have created a binary network (using I-graph in R) then applied "Cluster_optimum" function (Igraph) to cluster them. For certain range of cutoff value, R stopped working.

Now, I have found TramineR is a package that is used for sequence clustering. Is the package used for only state clustering (A,B,c..G) or it can be used in my case (like which DF has most similar sequence)?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.