multiview embedding forecast
Usage
# S4 method for class 'sf'
multiview(
data,
columns,
target,
nvar,
lib = NULL,
pred = NULL,
E = 3,
tau = 1,
k = E + 2,
nb = NULL,
top = NULL,
threads = detectThreads(),
trend.rm = TRUE
)
# S4 method for class 'SpatRaster'
multiview(
data,
columns,
target,
nvar,
lib = NULL,
pred = NULL,
E = 3,
tau = 1,
k = E + 2,
top = NULL,
threads = detectThreads(),
trend.rm = TRUE
)Arguments
- data
The observation data.
- columns
Names of individual variables.
- target
Name of target variable.
- nvar
Number of variable combinations.
- lib
(optional) Libraries indices.
- pred
(optional) Predictions indices.
- E
(optional) Dimensions of the embedding.
- tau
(optional) Step of spatial lags.
- k
(optional) Number of nearest neighbors used for prediction.
- nb
(optional) The neighbours list.
- top
(optional) Number of reconstructions used for MVE forecast.
- threads
(optional) Number of threads.
- trend.rm
(optional) Whether to remove the linear trend.
References
Ye H., and G. Sugihara, 2016. Information leverage in interconnected ecosystems: Overcoming the curse of dimensionality. Science 353:922-925.
Examples
columbus = sf::read_sf(system.file("case/columbus.gpkg", package="spEDM"))
# \donttest{
multiview(columbus,
columns = c("inc","crime","open","plumb","discbd"),
target = "hoval", nvar = 3)
#> [1] 18.8646054 22.4869617 21.1550738 -1.1253460 -9.1007114 -2.7903615
#> [7] -7.6201012 -6.6613986 -8.7202885 16.2907046 -13.4100870 5.5617166
#> [13] -4.6242104 -6.0215133 -12.1983725 -9.3303243 14.4441917 2.2055448
#> [19] -4.3641292 -6.0871013 -11.5193470 2.6235111 13.7585733 -8.7467063
#> [25] -9.7543853 -4.1516025 3.8137697 -18.6397054 -12.1462767 -12.0658493
#> [31] 10.9218398 19.4799373 -8.1531852 16.5727910 0.7379578 8.5333685
#> [37] -15.5058094 -17.8177107 23.9793164 27.1593848 18.1874514 -2.9723211
#> [43] -2.8697946 -8.5016198 1.2200571 5.7208429 15.1875343 6.5063233
#> [49] 0.2310141
# }