gtd_pr: Plot Precision and Recall scores

In maskedforreview/gtdnews: Online appendix for @@ paper

Description

Uses the gold standard in the gold_matches data to calculate Precision and Recall scores. Results are calculated separately for the article level (does the article have at least one GTD match) and match level (each individual match of an article to a GTD event). Results are also calculated for different Similarity thresholds, which will be plotted on the x-axis.

Usage

gtd_pr(
  g,
  weight_range,
  steps,
  weights = NULL,
  weight_col = "weight",
  filter = NULL,
  exclude_added = T,
  gtd_id_filter = NULL
)

Arguments

g	An edgelist, as created with newsflow.compare
weight_range	The range of weight to plot on the x axis
steps	The number of ticks on the x axis
weights	Optionally, provide a vector of weights. This overrides the weight_range and steps arguments
weight_col	the name of the column with the weight scores
filter	Optionally, a logical vector indicating which rows of g$d to use.

Details

The results are also returned as a list with data.frames. See gold_matches$description for details on the gold standard.

Value

A plot an a list with data.frames

maskedforreview/gtdnews documentation built on April 12, 2021, 11:53 a.m.