Skip to contents

Summarize an evaluation into overall and per-class metrics

Usage

eval_summary(results)

Arguments

results

Output of evaluate_columns().

Value

A list with overall (a one-row data frame: column counts, class and type accuracy over in-scope columns, and prediction coverage) and by_class (per gold class: support, recall, precision, f1). Precision is computed over ALL predictions of a class (in- and out-of-scope columns), so a detector firing on out-of-scope columns is penalized.