Fits flexible additive models to data on stream networks, taking account of flow-connectivity of the network. Models are fitted using penalised least squares.
CHANGES IN VERSION 2.1.1 30/03/2017
-----------------------------------
- updated network names in examples to avoid error due to attempt to overwrite a network
CHANGES IN VERSION 2.1 29/08/2016
---------------------------------
- network(max.df = ...) removed and replaced with network(fixed.df = ...), for fixing the
degrees of freedom in a network smooth.
- parameter combinations causing optimiser to encounter singular penalty matrices are now
avoided.
- fixed error with control = list(approx = 1)
- cyclical smooths via m(X, cyclical = T) and interactions via m(X, Y, cyclical = c(T, F))
- n-dimensional smooth accepted using m()
- removed ``network-node'' option for plot.smnet
- removed dependency on 'igraphs'
CHANGES IN VERSION 2.0 08/06/2015
---------------------------------
- get_adjacency and smnet reworked so that SSN objects with multiple networks supported.
get_adjacency now returns a list containing the adjacency matrix and the binary id table
from the SSN object.
- significant changes to plotting of spatial networks
- dependence on 'dfoptim' package not required, optimiser changed to 'optim()'
- show_weights() function added for exploring valid weights in data
- automatic Shreve weights, "autoShreve"", option added for 'weights' argument in network if
valid weights are not available
- examples updated