| Type: | Package |
| Title: | Locally Scaled Density Based Clustering |
| Version: | 0.1.0 |
| Author: | Fella Ulandari and Robert Kurniawan |
| Maintainer: | Fella Ulandari <16.9134@stis.ac.id> |
| Description: | Implementation of Locally Scaled Density Based Clustering (LSDBC) algorithm proposed by Bicici and Yuret (2007) <doi:10.1007/978-3-540-71618-1_82>. This package also contains some supporting functions such as betaCV() function and get_spectral() function. |
| License: | GPL-2 | GPL-3 [expanded from: GPL] |
| Encoding: | UTF-8 |
| LazyData: | true |
| Imports: | stats |
| RoxygenNote: | 7.1.0 |
| NeedsCompilation: | no |
| Packaged: | 2020-06-29 14:44:28 UTC; fella |
| Repository: | CRAN |
| Date/Publication: | 2020-06-29 15:30:13 UTC |
BetaCV
Description
function to calculates the BetaCV.
Usage
betaCV(clust,dist)
Arguments
clust |
Determine in which cluster a data is belonged. clust should be a numeric, 0 indicates a noise and cluster start at 1. |
dist |
Distance matrix |
Details
BetaCV measures how well the clusters based on compactness (intra-cluster distance) and separability (inter-cluster distance). BetaCV is the ratio between the average of intra-cluster distance to the average of inter-claster distance. The smaller BetaCV value indicates the better the clustering.
Value
This function returns the betaCV value.
Author(s)
Fella Ulandari and Robert Kurniawan
References
University of Illinois. (2020, January 10). 6.1 Methods for Clustering Validation. Retrieved from Coursera: https://www.coursera.org/lecture/cluster-analysis/6-1-methods-for-clustering-validation-k59pn
See Also
https://www.coursera.org/lecture/cluster-analysis/6-1-methods-for-clustering-validation-k59pn
Examples
x <- runif(20,-1,1)
y <- runif(20,-1,1)
dataset <- cbind(x,y)
l <- lsdbc(dataset, 7,3,"euclidean")
dmat <- as.matrix(dist(dataset,"euclidean"))
betaCV(l$cluster,dmat)
Generate Spectral Data
Description
Generate a dataset with spectral distribution.
Usage
get_spectral(n)
Arguments
n |
Number of data to be generated |
Value
This function returns a dataframe with the spectral distribution
Author(s)
Fella Ulandari and Robert Kurniawan
References
Bicici, E., & Yuret, D. (2007). Locally Scaled Density Based Clustering. International Conference on Adaptive and Natural Computing Algorithms (pp. 739-748). Berlin: Springer.
Examples
##Generate 1000 data##
get_spectral(1000)
Locally Scaled Density Based Clustering
Description
Generate a locally scaled density based clustering as proposed by Bicici and Yuret (2007).
Usage
lsdbc(data, k, alpha, jarak = c("euclidean", "manhattan", "canberra", "geodesic"))
Arguments
data |
Dataset consists of two variables (x,y) indicating coordinates of each data (point) |
k |
Number of neighbor to be considered |
alpha |
Parameter for determining local maximum |
jarak |
Type of distance to be used, the options are c("euclidean", "manhattan", "canberra", "geodesic") |
Value
This function returns a list with the following objects:
data |
a dataframe of the dataset used. |
cluster |
an integer vector coding cluster membership, 0 indicates a noise and cluster start at 1. |
parameter |
consist of parameter k and alpha. |
Author(s)
Fella Ulandari and Robert Kurniawan
References
Bicici, E., & Yuret, D. (2007). Locally Scaled Density Based Clustering. International Conference on Adaptive and Natural Computing Algorithms (pp. 739-748). Berlin: Springer.
See Also
https://doi.org/10.1007/978-3-540-71618-1_82
Examples
x <- runif(20,-1,1)
y <- runif(20,-1,1)
dataset <- cbind(x,y)
l <- lsdbc(dataset, 7,3,"euclidean")
l