Table 2 Average performance metrics of all models across monitoring nodes.

From: Ocean environment prediction methods based on deep learning and spatiotemporal feature fusion

Lead

Metrics

Mod1

Mod2

Mod3

Mod4

Mod5

Mod6

Mod7

MSSTN

day lead

IMAPE (%)

4.402

2.453

2.465

2.224

1.748

1.719

1.807

1.678

IRMSLE

0.0784

0.0419

0.0366

0.0336

0.0328

0.0273

0.0238

0.0206

week lead

IMAPE (%)

4.709

2.702

2.575

2.394

2.203

2.108

2.116

2.077

IRMSLE

0.0834

0.0512

0.045

0.0411

0.0376

0.0316

0.0303

0.0267

  1. Metrics include IMAPE and IRMSLE. Lower values indicate better prediction accuracy. The table highlights that MSSTN consistently achieves the lowest error values, demonstrating superior predictive accuracy compared with statistical, temporal, and spatiotemporal approaches.
Back to article page