Table 3.
Confusion matrices showing the proportion of accurate predictions based on EM, FBG-sampling, and approximate sampling methods on 500 simulated datasets
Truth | ||||
---|---|---|---|---|
Loss | Neutral | Gain | ||
Loss | 0.855 | 0.071 | 0.074 | |
EM | Neutral | 0.001 | 0.996 | 0.003 |
Gain | 0.190 | 0.087 | 0.723 | |
Loss | 0.980 | 0.020 | 0.000 | |
FBG | Neutral | 0.002 | 0.995 | 0.003 |
Gain | 0.002 | 0.020 | 0.973 | |
Loss | 0.981 | 0.019 | 0.000 | |
Approx. (w = 1.25σD) | Neutral | 0.002 | 0.993 | 0.005 |
Gain | 0.009 | 0.022 | 0.969 |