Table 2. Barzooka screening tool performance in an external validation set of bioRxiv preprints.
Class | Manually labeled cases | False positives | Precision | Recall | F1 score |
---|---|---|---|---|---|
Bar graph of counts or proportions (appropriate) | 345 | 60 | 0.82 | 0.81 | 0.82 |
Bar graph of continuous data (inappropriate) | 405 | 25 | 0.94 | 0.92 | 0.93 |
Bar graph with dot plot | 74 | 37 | 0.63 | 0.86 | 0.73 |
Dot plot | 257 | 51 | 0.80 | 0.80 | 0.80 |
Box plot | 255 | 36 | 0.87 | 0.91 | 0.89 |
Violin plot | 57 | 27 | 0.65 | 0.89 | 0.76 |
Histogram | 198 | 66 | 0.72 | 0.85 | 0.78 |
Flow chart | 20 | 26 | 0.40 | 0.85 | 0.54 |
Pie chart | 71 | 12 | 0.83 | 0.85 | 0.84 |
The bioRxiv external validation set included 1107 bioRxiv preprints published in May 2019.