summary.transition: Obtain preference lists for predicted and actual (reference)...

In PAutilities: Streamline Physical Activity Research

Obtain preference lists for predicted and actual (reference) activity transitions

Description

When predicting activity transitions, the behavior of the predictor is not known a priori. It may predict too many or too few transitions, and its "intent" is also unknown. Therefore, some method is necessary in order to determine which predictions (if any) should be taken to correspond to a reference transition. There should also be a record of false positives and false negatives. The problem is treated as an instance of the college admissions problem, wherein both parties give their preferences for who they would like to be matched with, and a stable arrangement is sought. This function supports the overall goal by assigning the preferences based on the temporal proximity of predicted and actual transitions. Preferences beyond a specified window_size are not allowed.

Usage

get_preferences(predictions, references, window_size, missing_info)

## S3 method for class 'transition'
summary(object, ...)

Arguments

predictions	A dummy-coded vector of predicted transitions (1) interspersed with non-transitions (0). Logical vectors are coerced to numeric.
references	A dummy-coded vector of actual (i.e., reference) transitions (1) interspersed with non-transitions (0). Logical vectors are coerced to numeric.
window_size	The maximum number of indices that are allowed to separate a predicted and reference transition, before the two are considered incompatible
object	a transition object to analyze
...	further arguments passed to or from methods, currently unused

Value

A list of matrices giving distance-based preferences for both the predicted and reference transitions, formatted to pass directly to hri

a data frame containing indicators that reflect, in different ways, the effectiveness of predicted transitions compared to the set of actual (reference) transitions

Examples

predictions <- (sample(1:100)%%2)
references  <- (sample(1:100)%%2)
window_size <- 7
if (isTRUE(requireNamespace("matchingMarkets", quietly = TRUE))){
  transitions <- get_transition_info(
    predictions, references, window_size
  )
  summary(transitions)
}

PAutilities documentation built on Aug. 21, 2022, 9:05 a.m.