Table 3 Synthetic dataset linkage quality - estimated vs. calculated
From: Estimating parameters for probabilistic linkage of privacy-preserved datasets
Data Error Rate | Calculated Probabilities | EM m-probs and Estimated u-probs | ||||||
|---|---|---|---|---|---|---|---|---|
Highest | Estimated | Highest | Estimated | |||||
Threshold | FMeasure | Threshold | FMeasure | Threshold | FMeasure | Threshold | FMeasure | |
0% | 49 | 1.0000 | 8 | 0.9999 | 49 | 1.0000 | 8 | 0.9999 |
1% | 9 | 0.9979 | 16 | 0.9978 | 13 | 0.9979 | 11 | 0.9979 |
5% | 8 | 0.9549 | 16 | 0.9541 | 12 | 0.9549 | 11 | 0.9549 |
10% | 8 | 0.8443 | 16 | 0.8399 | 12 | 0.8439 | 11 | 0.8436 |
20% | 8 | 0.5217 | 16 | 0.4938 | 12 | 0.4999 | 11 | 0.4917 |