CODEWITHSUNDEEP
matlabmatrixsparse-matrixmarkov-chains
LarsVegas
LarsVegas

Reputation: 65

Markov Chain Transition Matrix: MATLAB function sparse - index exceeds matrix dimensions

I'm referencing to this Question: Estimate Markov Chain Transition Matrix in MATLAB With Different State Sequence Lengths

The described procedure worked out perfectly for me, but I'm not able to adapt the last Matlab command to create the transition matrix S = sparse(i,j,v,m,n)

transitionMatrix = sparse(uniqueTransitions(:,1), uniqueTransitions(:,2), p, 6,6)

Here is my data with 18 unique Transitions and the corresponding probabilities:

uniqueTransitions       p
5405    5414            0,046511628
5414    5420            0,046511628
5420    5420            0,139534884
5420    9999            0,046511628
5414    5430            0,046511628
5430    5414            0,023255814
5430    5191            0,023255814
5191    5462            0,023255814
5462    5414            0,023255814
5414    5414            0,139534884
5414    9999            0,023255814
5405    5419            0,023255814
5419    5419            0,162790698
5419    5412            0,069767442
5412    5419            0,046511628
5412    5414            0,023255814
5405    5405            0,046511628
5405    9999            0,046511628

How to calculate the parameters 'm' and 'n'? I already read the online Matlab documentation and tried a lot of parameters but I can't find a solution.

Upvotes: 0

Views: 331

Answers (1)

Daniel
Daniel

Reputation: 36710

You are creating a sparse matrix of size 6 by 6 but your data is obviously larger. Assuming each state is reachable you can simple skip the size argument:

transitionMatrix = sparse(uniqueTransitions(:,1), uniqueTransitions(:,2), p)

It is only important when you want to create a matrix with last row or last column containing only zeros.

Upvotes: 1

Related Questions