cec() now accepts an optional weights
argument: a numeric vector of non-negative values, one per observation.
Cluster means, covariances, and the returned $probability
field are all weight-aware. When weights = NULL (default),
behaviour is identical to previous versions.weights is non-uniform, $probability
now reports the weighted mixing proportion W_k / W_total
rather than the observation-count fraction n_k / m.weights is non-uniform, card.min is
now interpreted as a minimum weight sum rather than a minimum
observation count.weights: a cluster with
fewer observations than dimensions could pass the weight-sum-based
card.min check and produce a singular covariance. Cluster
removal now also enforces a minimum observation count.cec(..., keep.removed = TRUE) whenever
a cluster was actually removed during clustering
(Error in matrix(NA, 1, ncol(center)) : non-numeric matrix extent).plot.cec() now warns and skips a single cluster’s
covariance ellipse if it cannot be drawn (e.g. a degenerate covariance),
instead of failing the whole plot. Previously the surrounding
tryCatch had no error handler and so provided
no protection.Initial Release.
Need a high-speed mirror for your open-source project?
Contact our mirror admin team at info@clientvps.com.
This archive is provided as a free public service to the community.
Proudly supported by infrastructure from VPSPulse , RxServers , BuyNumber , UnitVPS , OffshoreName and secure payment technology by ArionPay.