Prepare land suitability dataset to model

Load packages, read data and source custom scripts

rm(list = ls())
library(ggplot2)
library(scales)
library(stars)

#> Loading required package: abind

#> Loading required package: sf

#> Linking to GEOS 3.8.0, GDAL 3.0.4, PROJ 6.3.1

path_proj <- day2day::git_path()
path_data <- file.path(path_proj, "data")
path_processed <- file.path(path_data, "processed")

land_df <- fst::read_fst(file.path(path_processed, "uk_1km_dataframe.fst"))

Prepare input data for modelling

Remove all the rows with at least one missing value.

land_df <- na.omit(land_df)

Visualize full dataset for modelling

Visualize the pixels that will be modelled.

ggplot(land_df, aes(x, y)) +
    geom_tile(aes(fill = factor(country))) +
    labs(x = NULL, y = NULL, color = NULL, fill = "Country") +
    theme_bw() +
    coord_fixed() +
    scale_x_continuous(labels = label_number(scale = 1 / 1e3, suffix = " km")) +
    scale_y_continuous(labels = label_number(scale = 1 / 1e3, suffix = " km"))

Pseudo mask-out urban

quantile(land_df$count_6_urban, seq(0.45, 1, 0.05))

#>  45%  50%  55%  60%  65%  70%  75%  80%  85%  90%  95% 100% 
#>    0    2   10   14   22   32   49   81  147  313  813 1600

land_df_maskout <- subset(land_df, count_6_urban < quantile(count_6_urban, 0.9))

Visualize masked out dataset for modelling

ggplot(land_df_maskout, aes(x, y)) +
    geom_tile(aes(fill = factor(country))) +
    labs(x = NULL, y = NULL, color = NULL, fill = "Country") +
    theme_bw() +
    coord_fixed() +
    scale_x_continuous(labels = label_number(scale = 1 / 1e3, suffix = " km")) +
    scale_y_continuous(labels = label_number(scale = 1 / 1e3, suffix = " km"))

Write data to be modelled

fst::write_fst(land_df, file.path(path_processed, "uk_1km_dataframe_train_full.fst"))
fst::write_fst(land_df_maskout, file.path(path_processed, "uk_1km_dataframe_train_maskout.fst"))

Time to execute the task

Only useful when executed with Rscript.

proc.time()

#>    user  system elapsed 
#>  56.876   1.765  58.559