This is a modified kmeans clustering technique to automatize the number of groups or clusters that can be partitioned the sample. Several techniques are used to obtain the best number of clusters.

Clustering(data, n = "auto", n_max = 10, iter.max = 10,
  auto_criterion = c("explainwss", "db", "ratkowsky", "ball",
  "friedman"), confidenceWSS = 0.9, agregate_method = median)

Arguments

data

Data frame which numeric variables.

n

Data frame which numeric variables.

n_max

maximal number of clusters, between 2 and (number of objects - 1), greater or equal to n_min. By default, n_max=10.

iter.max

the maximum number of iterations allowed.

auto_criterion

the available criterions are: "explainwss", "db", "ratkowsky", "ball" and "friedman".

confidenceWSS

a confidence interval for criterion WSS.

agregate_method

a function to agregate results of different methods. Default value=median

Details

Several methods are available in order to obtain the best number of clusters: explainwss = Within-cluster Sum of Square db = Davies–Bouldin index (DBI). Davies and Bouldin (1979) ratkowsky = Ratkowsky and Lance (1978) ball = Ball and Hall (1965) friedman = Friedman and Rubin (1967)

@return A MLA object of subclass Clustering

Examples

## Load a Dataset
# NOT RUN { data(EGATUR) modelFit <- Clustering(data=EGATUR[,c("A13","gastototal")]) # }