Fig. 10: Coverage rates, defined as the proportion of test cases where at least one of the predicted scenarios closely matched the actual drifting pattern (within the top 5% of smallest Dpum), as the function of rank k. | npj Natural Hazards

Fig. 10: Coverage rates, defined as the proportion of test cases where at least one of the predicted scenarios closely matched the actual drifting pattern (within the top 5% of smallest Dpum), as the function of rank k.

From: Data-driven proactive prediction of pumice drifting patterns using similarity search of the Kuroshio current axis

Fig. 10

Higher coverage rates indicate better model performance in identifying realistic scenarios. Plotted are the results of each KCA distance metric and the baseline results of random selections without a reference (black broken line).

Back to article page