is binning technique will reduce quality of hmmtrain function?

I am training hmm with the following code.

[ESTTR,ESTEMIT] = hmmtrain(seq,TRGUESS,EMITGUESS)

Here my 'seq' is a matrix with floating values. Since hmmtrain is not accepting floating sequences, i have applied binning technique on my seq matrix. All the floating values in my sequences are replaced with corresponding bin indexes. Is this binning technique will reduce the accuracy of model?

Another issue is, after converting seq matrix into bin indexes, i was getting the following error message.

"SEQ must consist of integers between 1 and 25", because one of the value in my seq is 25. This 25 value is generated in sequence because i have chosen my number of bins 25. Therefore i have bins between 1 to 25 and corresponding indexes.

My question is, what is N represents in the error message " * SEQ must consist of integers between 1 and N*", and what are the limits of N. cant i use any symbol in hmmtrain?

Answer 1

[Jonathan Kwang]

In the documentation for "hmmtrain", N is described as "the number of possible emissions". https://www.mathworks.com/help/stats/hmmtrain.html