recommend: Function for recommending the modelled probabilities to use.
In sumitrahman/perfman: Probability Estimates From Ordered Logistic Regression

Description Usage Arguments Details Value Examples

This function recommends the choice of modelled probabilities for the chosen variable.

1	recommend(variable, lower.bound = -100, upper.bound = 100)

variable

A column name from the original data frame. This will be the variable one of whose values you are calculating the probability for.

This function selects the member of the family of probability distributions that are consistent with the model fitted by fit.model which is closest to the overall observed distribution in the original dataset. The resulting probabilities are what we call corrected probabilities and can be thought of as the probability distributions for each value of the chosen variable having corrected for the confounding effects of the other variables in the model.

There is text output displaying the recommended value of X - this is the value that is used to generate the specific probabilities from the family of probability distributions that are consistent with the logisitc effects for the variable of interest in the fitted model. Then a data frame is created (and displayed) that gives the probability distribution for each value of the variable of interest, and the implied distribution of outcomes (which is the mean of the various probability distributions, weighted by the relative frequencies of each particular value of the variable in the original dataset).

A data frame called optimised.probs that gives the recommended probabilities for each outcome and each value of the chosen variable, as well as the implied distribution for outcomes given these recommended probabilities.