Table 3 Base-table 26 from the original paper showing the counts and distances in meters (median and IQR) for the original data and the three anonymization methods6.

From: Comparison of Three Anonymization Tools for a Health Fitness Study

Commuting group

From home to school

From school to home

N (%)

Distance (IQR)

N (%)

Distance (IQR)

Car

170 (24%) 182 (26%) 177 (25%) 169 (24%)

3133 (3945) 3758 (3915) 7602 (8467)3754 (4215)

71 (10%) 58(8%)95(13%) 70 (10%)

3615 (3896) 3910 (3800) 3934 (7362)2584 (3560)

Public

230 (32%) 212 (30%) 221 (31%) 232 (33%)

4782 (4296) 4973 (4193) 5690 (8320) 4712 (4228)

245 (34%) 252 (35%) 210 (29%) 245 (34%)

4996 (4033) 5140 (3686) 2249 (5174) 5011 (4097)

Wheels

34 (5%) 31 (4%) 96 (13%) 36 (5%)

1366 (2211) 1356 (1378)6671 (8472)1094 (2251)

28 (4%) 14(2%)97(14%) 30 (4%)

1444 (2369) 2235 (3245)2741 (5282) 1337 (1376)

Walk

279 (39%) 278 (39%) 219(31%) 279 (39%)

799 (789) 805 (795) 5498 (8697) 787 (737)

369 (52%) 379 (53%) 311(44%) 368 (52%)

973 (1043) 954 (1062) 2374 (6068) 960 (1037)

Total

713 (100%) 703 (99%) 713 (100%) 716 (100%)

 

713 (100%) 703 (99%) 713 (100%) 713 (100%)

 
  1. Each group of four presents the data in order of Original, ARX, SDV, and SynDiffix. The bold/italic fonts for counts (N) are as described for Table 2. Distance and IRQ are bold font when the relative error is greater than 30%, and italics when the relative error is greater than 15%. Note that the original distances median and IQR don’t perfectly match those of the base-table 26 because of differences in the way median and IQR were calculated (Python versus R).
Back to article page