get_variable_effects: Create predicitons for all variables in a formula

In jacobaro/danalyze: Danalyze: Automated Analysis Functions

Create predicitons for all variables in a formula

Description

This function generates a list of predictions based on the data for all variables in the 'variables' formula. This is an alternative to showing coefficient estimates. Coefficient estimates can be hard to interpret or compare. This function produces standardized effect estimates for all variables.

Usage

get_variable_effects(
  variables,
  data,
  labels = NULL,
  .variable.set = NULL,
  .perc = c(0.1, 0.9),
  .few.values = 2/3
)

Arguments

variables	A formula or list of formulas that indicate which variables are included in summary statistics.
data	Dataframe with variables to produce summary statistics from.
labels	An optional vector of labels: c("variable" = "label).
.variable.set	An optional subset of variables to examine.
.perc	For numeric variables this sets the low and high quantile values. The default is to examine the effect of moving from the 10th percentile (0.1) to the 90th percentile (0.9).
.few.values	A value indicating the cut-off for dropping a value in the quantile. If one value represents more than this portion of the variable (e.g., more than 66 dropped.

Details

For character (factor) variables, all unique values of the variable are compared against each other. For numeric variables, the impact of moving from the 10th to the 90th percentile is estimated.

Value

A dataframe that shows the impact of all variables (median, confidence interval, and P-value). If labels are provided this dataframe will also contain the labels.

See Also

Other post analysis exploration: get_summary_statistics(), identify_examples()

jacobaro/danalyze documentation built on Oct. 20, 2022, 8:09 a.m.