regress: Regression intercept, slope, r2, and t-value across each set...

In adamlilith/fasterRaster: Faster Raster and Spatial Vector Processing Using 'GRASS GIS'

Regression intercept, slope, r2, and t-value across each set of cells

Description

This function performs a regression on each set of cells in a multi-layered GRaster. The output is a GRaster with the intercept, slope, r^2 value, and Student's t value. The regression formula is as y ~ 1 + x, where x is the layer number of each layer (e.g., 1 for the first or top layer in the input GRaster, 2 for the second or second-to-top layer, etc.). Note that this is restricted version of the functionality in terra::regress().

Usage

## S4 method for signature 'GRaster,missing'
regress(y, x, na.rm = FALSE)

Arguments

y	A multi-layer GRaster.
x	Ignored.
na.rm	Logical: If FALSE, any series of cells with NA in at least one cell results in an NA in the output.

Value

A multi-layer GRaster.

See Also

terra::regress()

Examples

if (grassStarted()) {

# Setup
library(sf)
library(terra)

# Example data
madChelsa <- fastData("madChelsa")

# Convert a SpatRaster to a GRaster
chelsa <- fast(madChelsa)
chelsa # 4 layers

# Central tendency
mean(chelsa)
mmode(chelsa)
median(chelsa)

# Statistics
nunique(chelsa)
sum(chelsa)
count(chelsa)
min(chelsa)
max(chelsa)
range(chelsa)
skewness(chelsa)
kurtosis(chelsa)

stdev(chelsa)
stdev(chelsa, pop = FALSE)
var(chelsa)
varpop(chelsa)

# Which layers have maximum/minimum?
which.min(chelsa)
which.max(chelsa)

# Regression

# Note the intercept is different for fasterRaster::regress().
regress(chelsa)
regress(madChelsa, 1:nlyr(madChelsa))

# Note: To get quantiles for each layer, use global().
quantile(chelsa, 0.1)

# NAs
madForest2000 <- fastData("madForest2000")
forest2000 <- fast(madForest2000)
forest2000 <- project(forest2000, chelsa, method = "near")

chelsaForest <- c(chelsa, forest2000)

nas <- anyNA(chelsaForest)
plot(nas)

allNas <- allNA(chelsaForest)
plot(allNas)

}

adamlilith/fasterRaster documentation built on Dec. 21, 2024, 2:04 a.m.