MIAmaxent (version 0.4.0)

plotResp2: Plot marginal-effect model response.

Description

plotResp2 plots the marginal-effect response of a given Maxent model over any of the included explanatory variables (EVs) in that model. For categorical variables, a bar plot is returned rather than a scatter plot. plotResp2 also returns a data frame containing the plotted data (for customizable graphics). Marginal-effect response curves present the response of the model when all other explanatory variables are held constant at their mean values (cf. single-effect response curves; plotResp).

Usage

plotResp2(data, EV, transformation, model, logscale = FALSE, ...)

Arguments

data

Data frame of explanatory variables (EVs) included in the model, with column names matching EV names. See readData.

EV

Name or column index of the explanatory variable in data for which the response curve is to be generated.

transformation

Full pathway of the 'transformations.Rdata' file containing the transformations used to build the model. This file is saved as a result of the deriveVars function. Equivalently, the second item in the list returned by deriveVars can be used directly.

model

Full pathway of the '.lambdas' file of the model in question. This file is saved as a result of selectEV.

logscale

Logical. Plot the common logarithm of PRO rather than PRO itself.

...

Arguments to be passed to plot to control the appearance of the plot. For example:

• cex for size of points
• col for color
• pch for type

Value

In addition to the graphical output, the plotted data is returned. In the case of a continuous EV, the plotted data is a list of 2:

1. respPts. Model response across individual data points. Columns in this data frame represent the following: EV value ("EV"), Probability Ratio Output of the model ("PRO"), and corresponding EV interval ("int").
2. respLine. Model response across intervals of the EV. Columns in this data frame represent the following: EV interval ("int"), number of points in the interval ("n"), mean EV value of the points in the interval ("intEV"), mean Probability Ratio Output of the points in the interval ("intPRO"), and exponentially weighted moving average of intPRO ("smoothPRO").

In the case of a categorical EV, the plotted data is a data frame containing the number of points in the level ("n"), the level name ("level"), and the mean Probability Ratio Output of the level ("levelRV").

Details

The plot contains points, representing the model response across individual data points, as well as a line, representing an exponentially weighted moving average of the model response over intervals of the EV. Model response is commonly plotted across EV values of the training data, but it is possible to plot the model response over any EV values supplied in data. The EV specified in data must not be an interaction term.

Examples

## Not run: ------------------------------------
# responseEV1 <- plotResp2(dat, "EV1",
#    transformation = "D:/path/to/modeling/directory/deriveVars/transformations.Rdata",
#    model = "D:/path/to/modeling/directory/selectEV/round/model/1.lambdas")
## ---------------------------------------------

names(toydata_selevs$selectedEV)
resp <- plotResp2(toydata_sp1po, "EV11", toydata_dvs$transformations,
   system.file("extdata/sommerfeltia", "1.lambdas", package = "MIAmaxent"))

## Not run: ------------------------------------
# # From vignette:
# pr_bygallResp2 <- plotResp2(grasslandPO, "pr_bygall",
# transformation = grasslandDVs[[2]],
# model = system.file("extdata", "1.lambdas", package = "MIAmaxent"))
## ---------------------------------------------