Skip to main content

Table 3 DP compared to SaTScanâ„¢ in terms of area under the ROC curve

From: Spatial cluster detection using dynamic programming

Frequency range

Coverage

DP AUC

SaTScanâ„¢

(DP-SaTScanâ„¢) 95% confidence interval limits

p-value

Low

< 16%

0.95047

0.49570

0.41018

0.49936

< 0.0001

 

16-34%

0.86260

0.46070

0.35816

0.44562

< 0.0001

 

≥ 35%

0.97409

0.42947

0.50375

0.58549

< 0.0001

Mid

< 16%

0.96019

0.65946

0.26789

0.33356

< 0.0001

 

16-34%

0.98246

0.85287

0.11099

0.14818

< 0.0001

 

≥ 35%

1.00000

0.81147

   

High

< 16%

0.98468

0.82496

0.14089

0.17856

< 0.0001

 

16-34%

0.99990

0.99490

0.00387

0.00611

< 0.0001

 

≥ 35%

1.00000

0.94950

   
  1. Comparison of the DP posteriors obtained from model averaging and SaTScanâ„¢ p-values in terms of ROC curves. The table shows the AUC for each algorithm, the lower and upper 95% confidence limits for the difference in areas (DP-SaTScanâ„¢) and the associated p-values