IMPROVING CONFIDENTIALITY WITH tau-ARGUS BY FOCUSSING ON CLEVER USAGE OF MICRODATA Roland van der...
-
Upload
tamsyn-murphy -
Category
Documents
-
view
215 -
download
1
Transcript of IMPROVING CONFIDENTIALITY WITH tau-ARGUS BY FOCUSSING ON CLEVER USAGE OF MICRODATA Roland van der...
IMPROVING CONFIDENTIALITY WITH tau-ARGUS BY FOCUSSING ON
CLEVER USAGE OF MICRODATA
Roland van der Meijden MSc.
± 10 minutes
Content of presentation
• tau-Argus• Tuning possibilities• Hierarchies• Historyfile, information loss and base material• Conclusions
Tau-Argus
Automated cell suppression software
Calculates confidentiality effects on all dimensions of a table simultaneously
Offers 4 confidentiality rules:- (n,k)-rule / dominance rule- p%-rule- p-q-rule / prior-posterior-rule- Minimum frequency rule
Tuning possibilities
Hierarchies: The way hierarchies are built is of influence on how secondary suppressions are applied.
History file: A preference can be given for which cells may or must be secondarily confidential.
Information loss weights: Information will be lost when applying secondary suppressions. The way tau-Argus calculates this information loss can be adjusted.
Base material: The way the microdata and preferred output are composed is of influence on the way secondary suppressions are applied.
Hierarchies (1)
Total Total Small
Large
A
B
C
D
E
0
1
2
3
4
5
6
7
0
1
2
3
4
5
6
7
Small
Large A
B
8
9
8
9
Old ‘narrow’ classification New ‘wide’ classification
Figure 1: A rearrangement of subcategories within a size class classification.
Hierarchies (2)
Status narrow size class
size class total
size class S - L
size class A - E
size class 1 - 9
% cells % cells % cells % cells
A frequency unsafe 32,8 42,0 52,9 61,4D secondary unsafe 27,2 31,5 25,1 17,9V safe 35,1 25,1 21,5 20,4
Status wide size class
size class total
size class S - L
size class A - E
size class 1 - 9
% cells % cells % cells % cells
A frequency unsafe 32,8 41,1 49,3 61,4D secondary unsafe 26,5 31,5 28,4 18,1V safe 35,7 25,8 21,6 20,2
Historyfile, information loss and base material (1)
Historyfile– Confidential, publishable, preferably do (not) suppress secondarily
Information loss– Cell value, frequency, equal and distance
Base material– Small area estimation, deliberately adjusting microdata and coordination of publication obligations
Historyfile, information loss and base material (2)
Methods for determining information loss
Cell value Frequency
Status2nd
digit NACE
3rd digit
NACE
4th digit
NACE
5th digit
NACE
2nd digit
NACE
3rd digit
NACE
4th digit NACE
5th digit
NACE
A frequency unsafe 0 195 2686 7995 0 195 2686 7995
B dominance unsafe 0 26 216 641 0 26 216 641
D secondary unsafe 4 321 2837 6806 4 298 2638 6423
V safe 285 1382 4813 8351 285 1405 5007 8724
Conclusions
- tau-Argus is a tool that is helpful in calculating confidentiality effects.
- The confidentiality pattern can be influenced.
- Improving the confidentiality pattern, takes a lot of effort.
- Both tooling and the way base material is used are of influence on the confidentiality pattern.