Table 5 The comparison results under different generation methods.

From: Predicting road traffic accident severity from imbalanced data using VAE attention and GCN

 

CHILI

original

ADASYN

VAE

GAN

DCGAN

Ours

 Accuracy

0.7041

0.6855

0.6982

0.7041

0.6893

0.8469

 Precision

0.5411

0.6876

0.4675

0.3013

0.1723

0.8606

 Recall

0.2872

0.6868

0.2728

0.2525

0.2500

0.8469

 F1-score

0.2816

0.6863

0.2546

0.2134

0.2040

0.8449

NEWYORK

 Accuracy

0.6853

0.5331

0.5816

0.5890

0.6318

0.8333

 Precision

0.4616

0.5369

0.2772

0.2954

0.3126

0.8399

 Recall

0.3705

0.5316

0.2713

0.2730

0.3352

0.8333

 F1-score

0.3565

0.4868

0.2433

0.2393

0.3199

0.8334

BRONX

 Accuracy

0.6232

0.4573

0.5945

0.5932

0.5931

0.7915

 Precision

0.3714

0.4664

0.1486

0.1483

0.1482

0.7993

 Recall

0.2911

0.4574

0.2500

0.2500

0.2500

0.7915

 F1-score

0.2685

0.4160

0.1860

0.1861

0.1872

0.7922

  1. Significant values are in bold.
Back to article page