Fig. 3: F1 scores of the networks constructed without GAM transformation (y-axis) plotted against the F1 scores of the GAM-transformed networks (x-axis) for all of the mock time-series datasets that were simulated. | ISME Communications

Fig. 3: F1 scores of the networks constructed without GAM transformation (y-axis) plotted against the F1 scores of the GAM-transformed networks (x-axis) for all of the mock time-series datasets that were simulated.

From: NetGAM: Using generalized additive models to improve the predictive power of ecological network analyses constructed using time-series data

Fig. 3: F1 scores of the networks constructed without GAM transformation (y-axis) plotted against the F1 scores of the GAM-transformed networks (x-axis) for all of the mock time-series datasets that were simulated.

A–C Show the comparison between the Glasso and the GAM-Glasso networks, while (D–F) show the comparison between the SCC and the GAM-SCC networks. The dashed, black lines show the 1:1 relationship. Data points below the 1:1 line depict network outputs that had a higher F1 score after applying the GAM-based data transformation, while data points that fall above the 1:1 line depict network runs that had a higher F1 score without applying the GAM-based data transformation prior to network construction.

Back to article page