Build an importance / performance model

Combines driver importance (against outcome) with feature performance (mean rating) into the tidy table consumed by plot_ipm(). Worded rating columns are recoded to 1-5 automatically.

Usage

ipm_model(
  data = NULL,
  outcome,
  rating_prefix,
  method = c("rwa", "forest", "correlation"),
  recode = TRUE,
  likert_levels = c("Very bad", "Bad", "Ok", "Good", "Very good")
)

Arguments

data

A data frame.

outcome

The outcome column (unquoted), e.g. nps_value.

rating_prefix

Column-name prefix identifying the rating block, e.g. "ratings_".

method

How to measure importance, passed to calc_importance(). Defaults to "rwa".

recode

If TRUE (default), character rating columns are mapped to 1-5 with recode_likert(); numeric columns are left as-is.

likert_levels

Scale wordings passed to recode_likert() when recoding.

Value

A tibble with feature, importance, performance and perf_class (the performance band as a 1-5 factor, from its integer part so a 3.57 mean sits in band 3, used for colouring).

Details

An importance/performance model answers two questions per feature at once: how much does it drive the outcome (importance, via calc_importance()) and how well are we doing on it (performance, the mean 1–5 rating). Plotting one against the other (plot_ipm()) reveals priorities: high-importance, low-performance features are where to invest. Worded rating columns are mapped to 1–5 with recode_likert() automatically (recode = TRUE); the prefix is stripped from feature names; and perf_class buckets performance into a 1–5 factor by its integer part (a 3.57 mean is band 3, not 4) for colouring. Importance comes from calc_importance(), so method chooses between relative weights (the default, requiring the suggested rwa package), a random forest and a plain correlation.

See also

calc_importance(), plot_ipm(), compare_values().

Other modelling: calc_importance(), calc_nps()

Examples

ipm_model(podracing_survey, nps_value, "ratings_")
#> Parsing `nps_value` as a non-binary variable.
#> Applying multiple regression to calculate relative weights...
#> # A tibble: 6 × 4
#>   feature    importance performance perf_class
#>   <chr>           <dbl>       <dbl> <fct>     
#> 1 value           24.9         2.86 2         
#> 2 atmosphere      23.8         3.57 3         
#> 3 speed           18.0         3.48 3         
#> 4 safety          14.1         2.88 2         
#> 5 venue           12.2         3.08 3         
#> 6 commentary       6.99        2.58 2