add alternative approach from 2013-01
[synclust.git] / man / findSyncVarRegions.Rd
... / ...
CommitLineData
1% Generated by roxygen2 (4.0.2): do not edit by hand
2\name{findSyncVarRegions}
3\alias{findSyncVarRegions}
4\title{Direct clustering from a neighborhoods graph, or get regions from (Poisson)
5distribution parameters optimization, using convex relaxation.}
6\usage{
7findSyncVarRegions(method, M, k, alpha, gmode, K, dtype, cmeth, pcoef = 1,
8 h = 0.001, eps = 0.001, maxit = 1000, showLL = TRUE, disp = TRUE)
9}
10\arguments{
11\item{method}{Global method: "direct" or "convex"}
12
13\item{M}{Matrix of observations in rows, the two last columns
14corresponding to geographic coordinates;
15set to NULL to use our initial dataset (625 rows / 9 years)}
16
17\item{k}{Number of neighbors}
18
19\item{alpha}{Weight parameter for intra-neighborhoods distance computations;
200 = take only geographic coordinates into account;
211 = take only observations over the years into account;
22in-between : several levels of compromise;
23-1 or any negative value : use a heuristic to choose alpha.}
24
25\item{gmode}{Neighborhood type. 0 = reduced [mutual] kNN; 1 = augmented kNN (symmetric);
262 = normal kNN; 3 = one NN in each quadrant; (NON-symmetric).
27NOTE: gmode==3 automatically sets k==4 (at most!)}
28
29\item{K}{Number of clusters}
30
31\item{dtype}{Distance type, in {"simple","spath","ectd"}.
32NOTE: better avoid "simple" if gmode>=2}
33
34\item{cmeth}{Clustering method, in {"KM","HC","spec"} for k-means (distances based)
35or hierarchical clustering, or spectral clustering (only if gmode>=2)}
36
37\item{pcoef}{Penalty value for convex optimization [default: 1.0]}
38
39\item{h}{Step in the min LL algorithm [default: 1e-3]}
40
41\item{eps}{Threshold to stop min.LL iterations [default: 1e-3]}
42
43\item{maxit}{Maximum number of iterations in the min LL algo [default: 1e3]}
44
45\item{showLL}{Print trace of log-likelihood evolution [default: true]}
46
47\item{disp}{True [default] for interactive display (otherwise nothing gets plotted)}
48}
49\value{
50list with the following entries. M: data matrix in input; NI: computed neighborhoods;
51 dists: computed distances matrix; clusts: partition into K clusters, as an integer vector;
52 cxpar: parameters obtained after convex optimization (if applicable)
53}
54\description{
55Direct clustering from a neighborhoods graph, or get regions from (Poisson)
56distribution parameters optimization, using convex relaxation.
57}
58\examples{
59cvr = findSyncVarRegions("convex",M=NULL,k=10,alpha=0.1,gmode=1,K=5,dtype="spath",cmeth="HC")
60drawMapWithSitez(cvr$M, cvr$clusters)
61drawNeighboroodGraph(cvr$M, cvr$NI)
62}
63