Perform DBSCAN clustering on functional data and optimize parameters `eps` and `minPts`.
Usage
fdbscan(
fdataobj,
eps,
minPts = 5,
metric = metric.lp,
par.metric = list(lp = 2)
)Arguments
- fdataobj
An object of class `fdata` containing functional data.
- eps
Neighborhood parameter (`eps`) for DBSCAN. If NULL, it is estimated automatically.
- minPts
Minimum cluster size (`minPts`). If NULL, it is estimated automatically.
- metric
Metric function to compute distances. Default is `metric.lp`.
- par.metric
List of arguments for the metric function.
Value
A list with the following elements:
`optimal`: Data frame with the best parameters (`eps`, `minPts`, `quality`).
`model`: DBSCAN clustering model with refined clusters.
`results`: Data frame with all combinations of `eps` and `minPts` tested.
Examples
if (FALSE) { # \dontrun{
t <- seq(0, 2 * pi, length.out = 101)
res <- rprocKclust(t, n = c(30, 50, 40), process = c("cos_sin", "sin", "cos"),
c = c(-1, 1, 1), k = c(NA, NA, NA), s = c(0.2, 0.3, 0.1))
opt_results <- optim.fdbscan(res$X, metric = metric.lp, par.metric = list(lp = 2))
print(opt_results$optimal)
plot(res$X, col = opt_results$model$cluster+1, main = "Optimal DBSCAN Clustering")
} # }