getPreds: Get model predictions

Description

This function allows getting the predictions of multiple models when applied to a given dataset. It can be useful if you have a list of model objects (e.g. resulting from multGLM) and want to apply them to a new data set containing the same variables for another region or time period. There are options to include the logit link ('Y') and/or 'Favourability' (see Fav).

Usage

getPreds(data, models, id.col = NULL, Y = FALSE, P = TRUE, 
Favourability = TRUE, incl.input = FALSE, verbosity = 2)

Value

This function returns the model predictions in an object of the same class as the input 'data', i.e. either a data frame or a RasterStack.

Arguments

data: an object of class either 'data.frame' or 'RasterStack' to which to apply the 'models' (below) to get their predictions; must contain all variables (with the same names, case-sensitive) included in any of the 'models'.
models: an object of class 'list' containing one or more model objects, obtained e.g. with function glm or multGLM.
id.col: optionally, the index number of a column of 'data' containing row identifiers, to be included in the result. Ignored if incl.input = TRUE, or if 'data' is a RasterStack rather than a data frame.
Y: logical, whether to include the logit link (y) value in the predictions.
P: logical, whether to include the probability value in the predictions.
Favourability: logical, whether to include Favourability in the predictions (see Fav).
incl.input: logical, whether to include input columns in the output data frame (if the 'data' input is a data frame too). The default is FALSE.
verbosity: numeric value indicating the amount of messages to display; currently meaningful values are 0, 1, and 2 (the default).

Author

A. Marcia Barbosa

Examples

Run this code

data(rotif.env)

names(rotif.env)


# identify rotifer data in the Eastern and Western hemispheres:

unique(rotif.env$CONTINENT)

rotif.env$HEMISPHERE <- "Eastern"

rotif.env$HEMISPHERE[rotif.env$CONTINENT %in%
c("NORTHERN_AMERICA", "SOUTHERN_AMERICA")] <- "Western"

head(rotif.env)


# separate the rotifer data into hemispheres

east.hem <- rotif.env[rotif.env$HEMISPHERE == "Eastern", ]
west.hem <- rotif.env[rotif.env$HEMISPHERE == "Western", ]


# make models for 3 of the species in rotif.env based on their distribution
# in the Eastern hemisphere:

mods <- multGLM(east.hem, sp.cols = 18:20, var.cols = 5:17, 
id.col = 1, step = FALSE, FDR = FALSE, trim = FALSE)


# get the models' predictions for the Western hemisphere dataset:

preds <- getPreds(west.hem, models = mods$models, P = TRUE, 
Favourability = TRUE)

head(preds)

Run the code above in your browser using DataLab

Data Engineering and BI courses are free this week!