Dawn: Executive function for testing AOA

In SchoenbergA/IKARUS: Classification Tool

Description

wrapper for IKARUS and AOA with buffered TrainPoints.

Usage

Dawn(
  FFS = FALSE,
  Tpoints,
  buf_size,
  design,
  Stk,
  Stk_name,
  plot_res = TRUE,
  save_png = FALSE,
  save_res = FALSE,
  path_res,
  path_png,
  fsize = 24
)

Arguments

FFS	boolean - if True uses a FFS. Default= FALSE
Tpoints	SpatialPoints - PointLayer with the TrainingPositions
buf_size	numeric - with for the buffer in meter
design	character - from for the buffer. Choose from "ROUND","FLAT","SQUARE"
Stk	RasterStack - with predictors for classfication
Stk_name	character - name of input Stack for the name to write in images and output name.
plot_res	boolean - set 'TRUE' to plot the prediction and AOA. Default = TRUE.
save_png	boolean - set 'TRUE' to write the images in.png format for prediction and AOA. Default = FALSE.
save_res	boolean - set 'TRUE' to write the prediction and AOA in .grd format. Default = FALSE.
path_res	character - the path to save the resulting RasterLayer. Required if "save_res" option is used.
fsize	numeric - the desired front size for the main and sub text. default = 24.
path_png_cl	character - the path to save the resulting plots for classification as png. Required if "save_png" option is used.
path_png_aoa	character - the path to save the resulting plots for AOA png. Required if "save_png" option is used.

Details

This function is a wrapper for the IKARUS workflow for a LLOCV Random Forest classification and further uses the AOA approach by Meyer 2020. By default the resulting prediction and AOA are plotted. By default further uses all varibales for the model, optional uses a FFS (if FFS=TRUE). Optional saves the RasterLayers. Further can save the results to .png format to a desired path.

Value

Returns the classification, model and the AOA. Optional saves the resuting RasterLayer and saves resuöting prediction and AOA to .png images.

Author(s)

Andreas Schönberg

Examples

# load libs
require(raster)
require(caret)
require(CAST)
require(doParallel)
require(rgeos)
# required for visualization
require(viridis)
require(png)
require(latticeExtra)
require(gridExtra)
# load data
lau_Stk <- raster::stack(system.file("extdata","lau_RGB.grd",package = "IKARUS"))
lau_tp <-rgdal::readOGR(system.file("extdata","lau_Tpoints.shp",package = "IKARUS"))
# handle CRS string
crs(lau_tp) <- crs(lau_Stk)
### run Dawn
# with training design 0.3 meter radius in cycles
test1 <- Dawn(FFS = F,Tpoints = lau_tp,buf_size = 0.3,design = "ROUND",Stk = lau_Stk,Stk_name = "RGB")
# access results
plot(test1$prediction) # prediciton
plot(test1$AOA) # AOA
test1$model_LLOCV # model
### Test series with DAWN
test1 <- Dawn(FFS = F,Tpoints = lau_tp,buf_size = 0.3,design = "ROUND",Stk = lau_Stk,Stk_name = "RGB")
test2 <- Dawn(FFS = F,Tpoints = lau_tp,buf_size = 0.6,design = "ROUND",Stk = lau_Stk,Stk_name = "RGB")
test3 <- Dawn(FFS = F,Tpoints = lau_tp,buf_size = 1.0,design = "ROUND",Stk = lau_Stk,Stk_name = "RGB")

SchoenbergA/IKARUS documentation built on Sept. 8, 2021, 11:11 a.m.