Description Usage Arguments Value
This function uses restricted_gibbs to arrive at a launch state and then propose a split-merge move. Based on the computed acceptance probability, this function accepts or rejects the proposal arriving at an old or new state for the cluster indicators.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 | split_merge(
Y,
z,
zs,
S,
mylist,
N,
t,
b,
log_v,
n_split,
Q,
p,
theta,
psi,
MORE_SPLIT = NULL,
partition_partial = NULL
)
|
Y |
multivariate binary data (row for subjects, column for dimensions) |
z |
original cluster indicators |
zs |
temporary variable used for split-merge assignments |
S |
temporary set of indicies for restricted Gibbs |
mylist |
the cluster ID list |
N |
a vector of the numbers of subjects in the list of clusters |
t |
total number of non-empty clusters |
b |
gamma parameter in the mixture of finite mixture formulation |
log_v |
coefficients in MFM (check Miller and Harrison 2016 JASA) |
n_split |
the number of intermediate Gibbs scan to arrive at the launch split state (usually 5) |
Q |
a Q matrix of dimension M by L |
p |
a vector of machine (factor) prevalences of length M |
theta |
a vector of true positive rates of length L |
psi |
a vector of false positive rates of length L |
MORE_SPLIT |
Default is |
partition_partial |
a list of subject ids that each belong to a few known clusters. |
Returns the values at the end of the current iteration
t
the number of clusters;
z
the cluster assignment indicators for all subjects (taking values from mylist
);
N
the number of subjects for each cluster; please refer to mylist to get the matching cluster IDs;
mylist
the non-empty and empty cluster IDs.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.