Skip to contents

Predict from a bag of models or from formula + coefficients

Source: R/bag_predict.R
predict.Rd

This function predicts responses for new data using either:

  • a bag object (via predict.bag), or

  • a model formula plus coefs and weights (via predict.formula).

Usage

predict(x, newdata, ...)

# S3 method for class 'bag'
predict(
  x,
  newdata,
  data = NULL,
  type = c("linear", "exponential", "exp", "logit", "cloglog")[1],
  wmean = TRUE,
  wq_probs = NULL,
  include = "all",
  baseline = c("median", "mean", "zero")[1],
  zoi = FALSE,
  zoi_shape = c("exp_decay", "gaussian_decay", "linear_decay", "threshold_decay")[1],
  which_cumulative = "cumulative",
  type_feature = c("point", "line", "area")[1],
  type_feature_recompute = FALSE,
  n_features = 1,
  zoi_limit = 0.05,
  resolution = 100,
  line_value = 1,
  ...
)

# S3 method for class 'formula'
predict(
  x,
  newdata,
  coefs,
  weights = 1,
  type = c("linear", "exponential", "exp", "logit", "cloglog")[1],
  wmean = TRUE,
  wq_probs = NULL,
  include = "all",
  ...
)

Arguments

x

[bag,list or formula]
A bag of models from bag_models(), or a formula used to build the model matrix for prediction.

newdata

[data.frame]
New data for prediction. Can contain all model variables or only focal variables.

...

[any]
Additional arguments passed to methods.

data

[data.frame=NULL]
Original data used for model fitting. Used in predict.bag() to recover baseline/categorical levels. Not used by predict.formula().

type

[character="linear"]{"linear", "exponential", "exp", "logit", "cloglog"}
Prediction scale.

wmean

[logical(1)=TRUE]
Whether to compute and return weighted mean prediction.

wq_probs

[numeric,vector=c(0.025, 0.5, 0.975)]
Weighted quantile probabilities for prediction summaries. If NULL, quantiles are not returned.

include

[character="all"]
Terms to include in prediction. Use "all" or one/more string patterns matching term names.

baseline

[character="median"]{"median", "mean", "zero"}
Baseline values for non-focal predictors.

zoi

[logical(1)=FALSE]
If TRUE, columns in newdata are treated as distance inputs and transformed into ZOI predictors.

zoi_shape

[character="exp_decay"]{"exp_decay", "gaussian_decay", "linear_decay", "threshold_decay"}
ZOI decay shape used when zoi = TRUE.

which_cumulative

[character(1)="cumulative"]
Pattern used to identify cumulative ZOI terms.

type_feature

[character="point"]{"point", "line", "area"}
Feature type for ZOI prediction.

type_feature_recompute

[logical(1)=FALSE]
Whether to recompute line- and area-feature geometry approximation for ZOI calculations.

n_features

[numeric(1)=1]
Number of features used in ZOI prediction.

zoi_limit

[numeric(1)=0.05]
Lower influence threshold used by non-vanishing ZOI decay functions (relevant for line-feature ZOI transformation). See zoi_functions().

resolution

[numeric(1)=100]
Raster resolution used for line-feature ZOI approximation. Used when recomputing the ZOI variables for line and area features.

line_value

[numeric(1)=1]
Value assigned to rasterized line cells when type_feature = "line". Used when recomputing the ZOI variables for line and area features.

coefs

[numeric vector or matrix]
Coefficients used by predict.formula(). If matrix, rows are term names and columns are model/resample coefficients.

weights

[numeric=1]
Model weights used by predict.formula() when combining predictions across models.

Value

A data.frame (or matrix-like object) with predicted values. Output columns depend on wmean and wq_probs:

  • weighted quantiles (if wq_probs is not NULL)

  • weighted mean (if wmean = TRUE)

  • individual model predictions (if wmean = FALSE and wq_probs = NULL)

Details

Predictions can be computed for full covariate data or for focal predictors only, while non-focal predictors are set to a baseline values (median, mean, or zero). It also supports ZOI-distance inputs that are internally transformed into ZOI predictors.

See also

plot_response(), create_linear_feature_zoi().

Examples

#---
# fit a bag to be tested

# load packages
library(glmnet)
library(ggplot2)

# load data
data("reindeer_rsf")
# rename it just for convenience
dat <- reindeer_rsf

# formula initial structure
f <- use ~ private_cabins_XXX + public_cabins_high_XXX +
  trails_XXX +
  NORUTreclass +
  # poly(norway_pca_klima_axis1, 2, raw = TRUE) +
  # poly(norway_pca_klima_axis2, 2, raw = TRUE) +
  norway_pca_klima_axis1 + norway_pca_klima_axis1_sq +
  norway_pca_klima_axis2 + norway_pca_klima_axis2_sq +
  norway_pca_klima_axis3 + norway_pca_klima_axis4

# add ZOI terms to the formula
zois <- c(100, 250, 500, 1000, 2500, 5000, 10000, 20000)
f <- add_zoi_formula(f, zoi_radius = zois, pattern = "XXX",
                     type = c("cumulative_exp_decay"),
                     separator = "", predictor_table = TRUE)$formula

# sampling - random sampling
set.seed(1234)
samples <- create_resamples(y = dat$use,
                            p = c(0.2, 0.2, 0.2),
                            times = 10,
                            colH0 = NULL)
#> [1] "Starting random sampling..."

# fit multiple models
fittedl <- bag_fit_net_logit(f,
                             data = dat,
                             samples = samples,
                             standardize = "internal", # glmnet does the standardization of covariates
                             metric = "AUC",
                             method = "AdaptiveLasso",
                             parallel = "mclapply",
                             mc.cores = 2)

# bag models in a single object
bag_object <- bag_models(fittedl, dat, score_threshold = 0.7)

#---
# prediction using formula

# new data, looking only at PCA1
dfvar = data.frame(norway_pca_klima_axis1 = seq(min(bag_object$data_summary$norway_pca_klima_axis1),
                                                max(bag_object$data_summary$norway_pca_klima_axis1),
                                                length.out = 100))
dfvar$norway_pca_klima_axis1_sq = dfvar$norway_pca_klima_axis1**2

# one model only
predict(x = f,
        newdata = dfvar,
        coefs = bag_object$coef[,1],
        include = "axis1")
#> Error in UseMethod("predict"): no applicable method for 'predict' applied to an object of class "formula"

# whole bag, weighted mean - here all weights = 1
predict(x = f,
        newdata = dfvar,
        coefs = bag_object$coef,
        include = names(dfvar))
#> Error in UseMethod("predict"): no applicable method for 'predict' applied to an object of class "formula"

# whole bag, for each model separately
bag_predict(x = f,
            newdata = dfvar,
            coefs = bag_object$coef,
            wmean = FALSE,
            include = names(dfvar))
#> Error in bag_predict(x = f, newdata = dfvar, coefs = bag_object$coef,     wmean = FALSE, include = names(dfvar)): could not find function "bag_predict"

#---
# prediction using bag

# prediction for the very same dataset, linear scale
predict(x = bag_object,
        newdata = dat,
        data = dat)
#> Error in UseMethod("predict"): no applicable method for 'predict' applied to an object of class "c('bag', 'list')"

# non ZOI variable
# new data, looking only at PCA3
dfvar = data.frame(norway_pca_klima_axis3 = seq(min(bag_object$data_summary$norway_pca_klima_axis3),
                                                max(bag_object$data_summary$norway_pca_klima_axis3),
                                                length.out = 100))

predict(x = bag_object,
        newdata = dfvar,
        data = dat)
#> Error in UseMethod("predict"): no applicable method for 'predict' applied to an object of class "c('bag', 'list')"

# ZOI variable
# new data, looking only at private cabins
dfvar = data.frame(private_cabins = 1e3*seq(0.2, 20, length.out = 100))

# prediction for 1 feature, linear scale
predict(x = bag_object,
        newdata = dfvar,
        data = dat,
        zoi = TRUE,
        baseline = "zero")
#> Error in UseMethod("predict"): no applicable method for 'predict' applied to an object of class "c('bag', 'list')"

# prediction for 30 features, exp scale, with weighted confidence intervals
predict(x = bag_object,
        newdata = dfvar,
        data = dat,
        type = "exp",
        wq_probs = c(0.025, 0.975),
        zoi = TRUE,
        n_features = 30,
        baseline = "zero")
#> Error in UseMethod("predict"): no applicable method for 'predict' applied to an object of class "c('bag', 'list')"

# plot
plot(dfvar[,1],
     predict(x = bag_object,
             newdata = dfvar,
             data = dat,
             type = "exp",
             zoi = TRUE,
             n_features = 30,
             baseline = "zero")[,1])
#> Error in h(simpleError(msg, call)): error in evaluating the argument 'y' in selecting a method for function 'plot': no applicable method for 'predict' applied to an object of class "c('bag', 'list')"