sscs-package: Search for structure in collective systems via redundancy...

In crtwomey/sscs: Searching for structure in collective systems

Description

An implementation of the clustering algorithm introduced in Twomey et al. (2018), provided as an R package. Currently only an implementation of the hard-clustering variant is provided; the plan is to include an improved implementation of the soft-clustering variant in a future release. The hard-clustering variant is by far the most practical for performance reasons, so it was made the focus of this initial release.

Details

Package:	sscs
Type:	Package
Version:	0.1
Date:	2019-08-28
License:	GPL-3

Author(s)

Colin Twomey

Maintainer: Colin Twomey <[e-mail]>

References

Twomey et al. (2018) Searching for structure in collective systems. https://doi.org/10.1101/362681

Examples

#
# A simple example using US state average temperature time
# series data, publicly available from NOAA here:
#     https://www.ncdc.noaa.gov/cag/statewide/time-series
#
# The results are intuitive and easy to visualize with a map.
#
## Not run: 
# univariate example
# ------------------

# create a new 'sscs' S3 clustering object
sscs <- new_sscs(US_state_temperature)

# run clustering
sscs <- run(sscs, nclusters=12, ncores=16, nreps=400)

# get the cluster assignments
cl <- assignments(sscs)

# convenience function for showing the result
plot_us_states_example(cl)


# multivariate example
# --------------------

# combine measurements
X <- cbind(US_state_temperature, US_state_precipitation)

# js describes how columns of X are grouped into
# multi-variate variables.
nstates <- ncol(US_state_temperature)
js      <- rep(1:nstates, 2)

# run clustering on states
sscs <- new_sscs(X, js)
sscs <- run(sscs, nclusters=12, ncores=16, nreps=400)
cl   <- assignments(sscs)

# show the result
plot_us_states_example(cl)


## End(Not run)

crtwomey/sscs documentation built on Dec. 25, 2019, 8:03 a.m.