Issue in Optimal_Clusters_GMM

[carversh]

Hi, I'm not sure exactly what the error is, however, I am using the Optimal_Clusters_GMM function and returning an error. When I do the same thing with the GMM function, everything works fine, however, I run into issues then optimizing.

Here is the function I'm using:
Optimal_Clusters_GMM(dat, max_clusters=10, criterion = 'BIC', dist_mode = 'maha_dist', seed_mode = 'random_subset', km_iter = 10, em_iter=10, var_floor = 1e-10, plot_data=T, verbose=T)

And here is the error message (note that all 10 iteration are completed:

ERROR while rich displaying an object: Error: $ operator is invalid for atomic vectors

Traceback:

1. tryCatch(withCallingHandlers({
 .     if (!mime %in% names(repr::mime2repr)) 
 .         stop("No repr_* for mimetype ", mime, " in repr::mime2repr")
 .     rpr <- repr::mime2repr[[mime]](obj)
 .     if (is.null(rpr)) 
 .         return(NULL)
 .     prepare_content(is.raw(rpr), rpr)
 . }, error = error_handler), error = outer_handler)
2. tryCatchList(expr, classes, parentenv, handlers)
3. tryCatchOne(expr, names, parentenv, handlers[[1L]])
4. doTryCatch(return(expr), name, parentenv, handler)
5. withCallingHandlers({
 .     if (!mime %in% names(repr::mime2repr)) 
 .         stop("No repr_* for mimetype ", mime, " in repr::mime2repr")
 .     rpr <- repr::mime2repr[[mime]](obj)
 .     if (is.null(rpr)) 
 .         return(NULL)
 .     prepare_content(is.raw(rpr), rpr)
 . }, error = error_handler)
6. repr::mime2repr[[mime]](obj)
7. repr_text.default(obj)
8. paste(capture.output(print(obj)), collapse = "\n")
9. capture.output(print(obj))
10. withVisible(...elt(i))
11. print(obj)
12. print.GMMCluster(obj)
13. cat("GMM Cluster\n", "Call:", deparse(x$call), "\n", "Data cols:", 
  .     ncol(x$centroids), "\n", "Centroids:", nrow(x$centroids), 
  .     "\n")
14. deparse(x$call)

[mlampros]

@carversh can you add a reproducible example? What is 'dat' that you use in 'Optimal_Clusters_GMM'?

[mlampros]

I can reproduce the error using the documentation example data and your parameter setting,

require(ClusterR)

data(dietary_survey_IBS)
dat = dietary_survey_IBS[, -ncol(dietary_survey_IBS)]
dat = center_scale(dat)

opt_gmm = Optimal_Clusters_GMM(data = dat, 
                               max_clusters = 10, 
                               criterion = "AIC",
                               dist_mode = 'maha_dist', 
                               seed_mode = 'random_subset',
                               km_iter = 10,
                               em_iter = 10,
                               var_floor = 1e-10, 
                               plot_data = TRUE,
                               verbose = TRUE)

opt_gmm
# Error: $ operator is invalid for atomic vectors

It seems that the error comes from the the following line

class(res) = c("GMMCluster", 'Gaussian Mixture Models')

if I modify this line to

class(res) =  'Gaussian Mixture Models'

then the function returns as expected. From what I understand it is related to "GMMCluster". The later appears in "predict.GMMCluster" and "print.GMMCluster"

@vspinu do you happen to know why this error appears and is related to "GMMCluster"? You've added this additional class in #24

Can we simply omit it here and keep only 'Gaussian Mixture Models'?

[vspinu]

I have added "GMMCluster" to all those models which where classed with the old 'Gaussian Mixture Models' and I didn't realize that this particular instance does not return a model. The documentation states that the returned value is a vector with either the AIC or BIC for each iteration. So it's not a model and that class should not even be there.

Let me PR to fix that.

[carversh]

Thanks so much!