Experimental consensus model decision

Treats the chi-squared deviance comparison, AIC, and BIC selections as three votes. This is an experimental relative model-selection summary; it does not replace the absolute goodness-of-fit assessment from assess_dose_response_models().

Usage

consensus_model_decision(object, alpha = 0.05)

Arguments

object

A comparison tibble returned by compare_dose_response_models(), a qdr_analysis, a qdr_model_set, or a list of two qdr_fit objects.

alpha

Significance level used when a comparison must be calculated.

Value

A tibble with one row per model and columns for each criterion's vote, total votes, consensus selection, agreement level, and conclusion.

Details

A model selected by all three criteria receives "unanimous" agreement. A model selected by two criteria receives "majority" agreement. Tied or unavailable criteria abstain, and fewer than two votes produces "no_consensus".