logitr (version 1.1.2)

predict.logitr: Predict probabilities and / or outcomes

Description

This method is used for computing predicted probabilities and / or outcomes for either the data used for model estimation or a new data set consisting of a single or multiple sets of alternatives.

Usage

# S3 method for logitr
predict(
  object,
  newdata = NULL,
  obsID = NULL,
  type = "prob",
  returnData = FALSE,
  interval = "none",
  level = 0.95,
  numDrawsCI = 10^4,
  pars = NULL,
  scalePar = NULL,
  randPars = NULL,
  randScale = NULL,
  ci,
  ...
)

Value

A data frame of predicted probabilities and / or outcomes.

Arguments

object

is an object of class logitr (a model estimated using the 'logitr()` function).

newdata

a data.frame. Each row is an alternative and each column an attribute corresponding to parameter names in the estimated model. Defaults to NULL, in which case predictions are made on the original data used to estimate the model.

obsID

The name of the column that identifies each set of alternatives in the data. Required if newdata != NULL. Defaults to NULL, in which case the value for obsID from the data in object is used.

type

A character vector defining what to predict: prob for probabilities, outcomes for outcomes. If you want both outputs, use c("prob", "outcome"). Outcomes are predicted randomly according to the predicted probabilities. Defaults to "prob".

returnData

If TRUE the data is also returned, otherwise only the predicted values ("prob" and / or "outcome") are returned. Defaults to FALSE.

interval

Type of interval calculation: "none" (default) or "confidence". Future versions will include "prediction" intervals as well.

level

Tolerance / confidence interval. Defaults to 0.95.

numDrawsCI

The number of draws to use in simulating uncertainty for the computed CI. Defaults to 10^4.

pars

The names of the parameters to be estimated in the model. Must be the same as the column names in the data argument. For WTP space models, do not include the scalePar variable in pars.

scalePar

The name of the column that identifies the scale variable, which is typically "price" for WTP space models, but could be any continuous variable, such as "time". Defaults to NULL.

randPars

A named vector whose names are the random parameters and values the distribution: 'n' for normal, 'ln' for log-normal, or 'cn' for zero-censored normal. Defaults to NULL.

randScale

The random distribution for the scale parameter: 'n' for normal, 'ln' for log-normal, or 'cn' for zero-censored normal. Only used for WTP space MXL models. Defaults to NULL.

ci

No longer used as of v1.1.0 - if provided, this is passed to the level argument, interval is set to "confidence", and a warning is displayed.

...

further arguments.

Examples

library(logitr)

# Estimate a preference space model
mnl_pref <- logitr(
  data    = yogurt,
  outcome = "choice",
  obsID   = "obsID",
  pars    = c("price", "feat", "brand")
)

# Predict probabilities and / or outcomes

# Predict probabilities for each alternative in the model data
probs <- predict(mnl_pref)
head(probs)

# Create a set of alternatives for which to make predictions.
# Each row is an alternative and each column an attribute.
data <- subset(
    yogurt, obsID %in% c(42, 13),
    select = c('obsID', 'alt', 'price', 'feat', 'brand'))
data

# Predict probabilities using the estimated model
predict(mnl_pref, newdata = data, obsID = "obsID")

# Predict probabilities and include a 95% confidence interval
predict(
  mnl_pref,
  newdata = data,
  obsID = "obsID",
  interval = "confidence",
  level = 0.95
)

# Predict outcomes
predict(mnl_pref, newdata = data, obsID = "obsID", type = "outcome")

# Predict outcomes and probabilities
predict(mnl_pref, newdata = data, obsID = "obsID", type = c("prob", "outcome"))