In suvedimukti/stdcab: Spatial Thinning, Dependency, Clustering, And Blocking of Point Data for Classification Problems

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

stdcab

The goal of the stdcab package is to apply spatial thinning on multi-class spatial point data, spatial cluster analysis for random and repeated cross-validations which partitions data into training and testing set compatible to tidy-modeling resamples. The package allows to run semivariance analysis, plotting and spatial blocking of observations.

Installation

You can install the released version of stdcab from CRAN with:

``` {r, installpkg, eval = TRUE } install.packages("stdcab")

To install the latest stdcab version from [GitHub](https://github.com/) use:

```r
remotes::install_github('suvedimukti/stdcab')

Repeated Cluster CV: rsample compatible

This function is extended based on spatial_cluster_sample, which is a wrapper around spatial_clustering_cv function of sptialsample package.

Load dataset

# load dataset
library(stdcab)

data(landcover)

# create another dataset based on landcover data that comes with the package
#
dspatial <- landcover

# examine data
head(dspatial)

# Class_name is the dependent (response ) data with seven classes (1 through  7)

Visualize the data

# load ggplot 2 for visualization
library(ggplot2)
ggplot(data = dspatial)+
  geom_sf(aes(colour = factor(Class_name)), size = 1.5, alpha = 0.8)+
  scale_colour_viridis_d()+
  labs(color = "Land Cover Class")+
  theme_bw(12)

Apply repeated cluster sampling on sf data

To make visualization easy lets make five folds and five repeats resulting 25 splits of data based on kmeans clustering.

# setting seeds
set.seed(1318)

spc_rcv <- repeated_spatial_cluster_sample(data = dspatial, v = 5, repeats = 5,
                            coords = NULL, spatial = TRUE, clust_method = "kmeans", 
                            dist_clust = NULL)

spc_rcv

Visualize clusters

Following chunk of code is a function to run each split at a time to visualize Analysis\Training and Assessment\Testing set in each fold and repeats.

library(magrittr) # 
fplot_splits <- function(split) {
  gp <- analysis(split) %>%
   dplyr::mutate(analysis = "Training") %>%
   dplyr::bind_rows(assessment(split) %>%
   dplyr::mutate(analysis = "Testing")) %>%
    ggplot(aes(X, Y, color = analysis)) +
    geom_point(alpha = 0.7, size = 2) +
    coord_fixed() +
    labs(color = "Train/Test") +
    scale_color_viridis_d(direction = -1) +
    xlab("Longitude (m)") +
    ylab("Latitidue (m)")
  print(gp)
}

Plotting

# plot using walk function from purrr package
# this requires gifski package
purrr::walk(spc_rcv$splits, fplot_splits)

suvedimukti/stdcab documentation built on Aug. 7, 2023, 2:28 p.m.