confidence_based_sensitivity_analysis.RdWhen estimating the effects of alternatives on the criteria, the raters can attach a confidence score to the estimates in the performance table (or, more accurately, to the entire performance table). This function uses these to test the robustness of the results.
confidence_based_sensitivity_analysis( multiEstimateDf, collapsedConfidences, scenarioDefinitions, weightProfiles, criteria, collapsedConfidences_criterionIdCol = "parentCriterion_id", transformationFunction = setToZero, scenarioOrder = names(scenarioDefinitions), scenarioLabels = stats::setNames(names(scenarioDefinitions), names(scenarioDefinitions)), scorer = "all", confidenceThresholds = seq(0, 1, by = 0.1), setMissingEstimates = 0, silent = mdmcda::opts$get("silent"), lineSize = 1, theme = ggplot2::theme_minimal(base_size = mdmcda::opts$get("ggBaseSize")) )
| multiEstimateDf | A multi estimate data frame that should contain columns `decision_id`, `decision_label`, `alternative_value`, `alternative_label`, `criterion_id`, `criterion_label`, and one or more estimates in columns named with the scorer identifiers. Columns with the `_id` suffix contain identifiers, and columns with the `_label` suffix contain human-readable labels. This dataframe is stored in the object called `multiEstimateDf` returned by a call to [read_performance_tables()] to read a set of scored performance tables. Note that different column names can be set using [mdmcda::opts]. |
|---|---|
| collapsedConfidences | The `collapsedConfidences` object. |
| scenarioDefinitions | The `scenarioDefinitions` object. |
| weightProfiles | The `weightProfiles` object. |
| criteria | The `criteria` object. |
| collapsedConfidences_criterionIdCol | The column name of the criterion identifiers in the `collapsedConfidences` object to use. |
| transformationFunction | The function to apply to the estimates in performance tables below the confidence threshold |
| scenarioOrder | The scenarios to process and the order in which to organise them. |
| scenarioLabels | The labels of the scenarios. |
| scorer | The scorer of whom to process the confidences. |
| confidenceThresholds | The confidence thresholds to process (by default, 0 to 1 in steps of .1). |
| setMissingEstimates | What to set missing values to, if any are encountered. |
| silent | Whether to suppress messages about progress. |
| lineSize | The line size to use in the plots. |
| theme | The `ggplot2` theme to use. |
An object with a data frame summarizing the sensitivity analysis, and two produced plots: one with the scores, and one with the ranks.
In the confidence-based sensitivity analyses, by default 10 estimates about which the raters expressed the least confidence are sequentially set to the result of the `transformationFunction`. The total score for each scenario is then computed again.