plot_idx {UniversalCVI}

Plots for visualizing CVIs

Description

Plot and compare upto 8 indices computed by the algorithms in this package.

Usage

plot_idx(idxresult,selected.idx = NULL)

Arguments

idxresult

a result from one of the algorithms FzzyCVIs, WP.IDX, GC.IDX, CCV.IDX, XB.IDX, WL.IDX, TANG.IDX, PBM.IDX, KWON.IDX, KWON2.IDX, KPBM.IDX, HF.IDX, Hvalid, Wvalid, SF.IDX, PB.IDX, DI.IDX, DB.IDX, CSL.IDX, CH.IDX or STRPBM.IDX.

selected.idx

a numeric vector indicates a part of the indexes from the idxresult in respective order selected by a user. For instance, selected.idx = 3 or selected.idx = c(1,3,5) may be selected. If not specified, the full idxresult will be considered.

Value

Plots of upto 8 cluster validity indices computed from FzzyCVIs, WP.IDX, GC.IDX, CCV.IDX, XB.IDX, WL.IDX, TANG.IDX, PBM.IDX, KWON.IDX, KWON2.IDX, KPBM.IDX, HF.IDX, Hvalid, Wvalid, SF.IDX, PB.IDX, DI.IDX, DB.IDX, CSL.IDX, CH.IDX or STRPBM.IDX. When using the isolated index algorithm, all the plots computed by that algorithm will be shown. When using FzzyCVIs or Hvalid with more than 8 selected indices, the first 8 indices will be plotted.

Author(s)

Nathakhun Wiroonsri and Onthada Preedasawakul

References

N. Wiroonsri, O. Preedasawakul, "A correlation-based fuzzy cluster validity index with secondary options detector," arXiv:2308.14785, 2023

See Also

FzzyCVIs, WP.IDX, XB.IDX, Hvalid

Examples

library(UniversalCVI)

# Iris data
x = iris[,1:4]

# ----Compute all the indices by FzzyCVIs ----
FCVIs = FzzyCVIs(scale(x), cmax = 10, cmin = 2, indexlist = 'all', corr = 'pearson',
                 method = 'FCM', fzm = 2, iter = 100, nstart = 20, NCstart = TRUE)

# plots of the eight indices by default
plot_idx(idxresult = FCVIs)

# plots of a specific selected.idx
plot_idx(idxresult = FCVIs, selected.idx = c(2,5,7))

# ----Compute all the indices by Wvalid ----
FCM.NC = Wvalid(scale(x), kmax = 10, kmin=2, method = 'kmeans',
  corr='pearson', nstart=100, NCstart = TRUE)

# plots of the four indices by default
plot_idx(idxresult = FCM.NC)

# ----Compute all the indices by XB.IDX ----

FCM.XB = XB.IDX(scale(x), cmax = 10, cmin = 2, method = "FCM",
  fzm = 2, nstart = 20, iter = 100)
plot_idx(idxresult = FCM.XB)