R/detect_envi_change.R
detect_envi_change.RdUse shapley values to detect the potential areas that will impact the species distribution. It only works on continuous variables.
detect_envi_change(
model,
var_occ,
variables,
target_var,
bins = NULL,
shap_nsim = 10,
seed = 10,
var_future = NULL,
variables_future = NULL,
pfun = .pfun_shap,
method = "gam",
formula = y ~ s(x)
)(isolation_forest or other model). It could
be the item model of POIsotree made by function isotree_po.
It also could be other user-fitted models as long as the pfun can work on it.
(data.frame, tibble) The data.frame style table that
include values of environmental variables at occurrence locations.
(stars) The stars of environmental variables.
It should have multiple attributes instead of dims.
If you have raster object instead, you
could use st_as_stars to convert it to stars or use
read_stars directly read source data as a stars.
You also could use item variables of POIsotree made by function
isotree_po.
(character) The selected variable to process.
(integer) The bin to cut the target variable for the analysis.
If it is NULL, no cut to apply. The default is NULL.
(integer) The number of Monte Carlo repetitions in SHAP
method to use for estimating each Shapley value. See details in documentation
of function explain in package fastshap.
When the number of variables is large, a smaller shap_nsim could be used.
Be cautious that making SHAP-based spatial dependence will be slow
because of Monte-Carlo computation for all pixels.
But it is worth the time because it is much more
informative. See details in documentation of function explain
in package fastshap. The default is 10. Usually a value 10 - 20 is enough.
(integer) The seed for any random progress. The default is 10L.
(numeric or stars) A number to apply to the current
variable or a stars layer as the future variable. It can be NULL if
variables_future is set.
(stars) A stars raster stack for future variables.
It could be NULL if var_future is set.
(function) The predict function that requires two arguments,
object and newdata.
It is only required when model is not isolation_forest.
The default is the wrapper function designed for iForest model in itsdm.
Argument passed on to geom_smooth to fit the line.
Note that the same arguments will be used for all target variables.
User could set variable one by one to set the arguments separately.
Default value is "gam".
Argument passed on to geom_smooth to fit the line.
Note that the same arguments will be used for all target variables.
User could set variable one by one to set the arguments separately.
The default is y ~ s(x).
(EnviChange) A list of
A figure of fitted variable curve
A map of variable contribiution change
Tipping points of variable contribution
A stars of variable contribution under current and future condition,
and the detected changes
The values show how changes in environmental variable affects the modeling prediction in space. These maps could help to answer questions of where will be affected by a changing variable.
# Using a pseudo presence-only occurrence dataset of
# virtual species provided in this package
library(dplyr)
library(sf)
library(stars)
library(itsdm)
#'
# Prepare data
data("occ_virtual_species")
obs_df <- occ_virtual_species %>% filter(usage == "train")
eval_df <- occ_virtual_species %>% filter(usage == "eval")
x_col <- "x"
y_col <- "y"
obs_col <- "observation"
#'
# Format the observations
obs_train_eval <- format_observation(
obs_df = obs_df, eval_df = eval_df,
x_col = x_col, y_col = y_col, obs_col = obs_col,
obs_type = "presence_only")
#'
env_vars <- system.file(
'extdata/bioclim_tanzania_10min.tif',
package = 'itsdm') %>% read_stars() %>%
slice('band', c(1, 12))
#'
# With imperfect_presence mode,
mod <- isotree_po(
obs_mode = "imperfect_presence",
obs = obs_train_eval$obs,
obs_ind_eval = obs_train_eval$eval,
variables = env_vars, ntrees = 5,
sample_size = 0.8, ndim = 1L,
nthreads = 1,
seed = 123L, response = FALSE,
spatial_response = FALSE,
check_variable = FALSE)
# Use a fixed value
bio1_changes <- detect_envi_change(
model = mod$model,
var_occ = mod$vars_train,
variables = mod$variables,
shap_nsim = 1,
target_var = "bio1",
var_future = 5)
if (FALSE) {
# Use a future layer
## Read the future Worldclim variables
future_vars <- system.file(
'extdata/future_bioclim_tanzania_10min.tif',
package = 'itsdm') %>% read_stars() %>%
split() %>% select(bioc1, bioc12)
# Rename the bands
names(future_vars) <- paste0("bio", c(1, 12))
## Just use the target future variable
climate_changes <- detect_envi_change(
model = mod$model,
var_occ = mod$vars_train,
variables = mod$variables,
shap_nsim = 1,
target_var = "bio1",
var_future = future_vars %>% select("bio1"))
## Use the whole future variable tack
bio12_changes <- detect_envi_change(
model = mod$model,
var_occ = mod$vars_train,
variables = mod$variables,
shap_nsim = 1,
target_var = "bio12",
variables_future = future_vars)
print(bio12_changes)
##### Use Random Forest model as an external model ########
library(randomForest)
# Prepare data
data("occ_virtual_species")
obs_df <- occ_virtual_species %>%
filter(usage == "train")
env_vars <- system.file(
'extdata/bioclim_tanzania_10min.tif',
package = 'itsdm') %>% read_stars() %>%
slice('band', c(1, 5, 12)) %>%
split()
model_data <- stars::st_extract(
env_vars, at = as.matrix(obs_df %>% select(x, y))) %>%
as.data.frame()
names(model_data) <- names(env_vars)
model_data <- model_data %>%
mutate(occ = obs_df[['observation']])
model_data$occ <- as.factor(model_data$occ)
mod_rf <- randomForest(
occ ~ .,
data = model_data,
ntree = 200)
pfun <- function(X.model, newdata) {
# for data.frame
predict(X.model, newdata, type = "prob")[, "1"]
}
# Use a fixed value
bio5_changes <- detect_envi_change(
model = mod_rf,
var_occ = model_data %>% select(-occ),
variables = env_vars,
target_var = "bio5",
bins = 20,
var_future = 5,
pfun = pfun)
plot(bio5_changes)
}