Geo-clustering for data de-identification

1. An apparatus, comprising:

a processor and memory configured to:

receive data at the processor representing a plurality of clusters, wherein each cluster represents a geographic area having a centroid and a population of individuals;

merge a first cluster having a smallest population with a second cluster having a nearest centroid based on a determination that the population of the first cluster is below a minimum size threshold to form an updated cluster that represents an updated geographic area having an updated centroid and an updated population that reflects the merging of the first cluster and the second cluster;

re-perform the merge until each cluster meets a minimum size threshold,

de-identify data records associated with the individuals of the clusters when the smallest cluster meets the minimum size threshold,

assess a risk of re-identification of the de-identified clusters based on k-anonymity,

increase the minimum size threshold and re-perform the merge, the de-identify, and the assess a risk, based on a determination that the assessed risk does not meet a risk criterion, and

present the de-identified clusters on a display based on a determination that the assessed risk meets the risk criterion.

2. The apparatus of claim 1 , wherein the processor and memory are further configured to:

identify a smallest one of the clusters based on a number of data records of the cluster; and

merge the smallest cluster with a second one of the clusters.

3. The apparatus of claim 2 , wherein the processor and memory are further configured to:

select the second cluster based on a geographic distance between geographic centroids of the smallest cluster and the second cluster.

4. The apparatus of claim 3 , wherein the processor and memory are further configured to:

select the second cluster based further on a socio-economic, a demographic, and/or a political parameter.

5. The apparatus of claim 1 , wherein the processor and memory are further configured to:

assess the risk of re-identification based further on optimal lattice anonymization.

6. The apparatus of claim 1 , wherein the processor and memory are further configured to:

retain a hierarchy of the merged clusters; and

utilize the hierarchy of the merged clusters with an optimal lattice anonymization process to automatically choose a generalization scheme to provide a k-anonymous dataset with minimal suppression.

7. The apparatus of claim 1 , wherein the update of the second cluster comprises calculating a new centroid based on the merged geographic areas of the first cluster and the second cluster.

8. A non-transitory computer readable medium encoded with a computer program that comprises instructions to cause a processor to:

receive data at the processor representing a plurality of clusters, wherein each cluster represents a geographic area having a centroid and a population of individuals;

merge a first cluster having a smallest population with a second cluster having a nearest centroid based on a determination that the population of the first cluster is below a minimum size threshold to form an updated cluster that represents an updated geographic area having an updated centroid and an updated population that reflects the merging of the first cluster and the second cluster;

re-perform the merge until each cluster meets a minimum size threshold;

de-identify data records associated with the individuals of the clusters when the smallest cluster meets the minimum size threshold,

assess a risk of re-identification of the de-identified clusters based on k-anonymity;

increase the minimum size threshold and re-perform the merge, the de-identify, and the assess a risk, based on a determination that the assessed risk does not meet a risk criterion; and

present the de-identified clusters on a display based on a determination that the assessed risk meets the risk criterion.

9. The non-transitory computer readable medium of claim 8 , further comprising instructions to cause the processor to:

identify a smallest one of the clusters based on a number of data records of the cluster; and

merge the smallest cluster with a second one of the clusters.

10. The non-transitory computer readable medium of claim 9 , further comprising instructions to cause the processor to:

select the second cluster based on a geographic distance between geographic centroids of the smallest cluster and the second cluster.

11. The non-transitory computer readable medium of claim 8 , further comprising instructions to cause the processor to:

select the second cluster based further on a socio-economic, a demographic, and/or a political parameter.

12. The non-transitory computer readable medium of claim 8 , further comprising instructions to cause the processor to:

assess the risk of re-identification based further on optimal lattice anonymization.

13. The non-transitory computer readable medium of claim 8 , further comprising instructions to cause the processor to:

retain a hierarchy of the merged clusters; and

utilize the hierarchy of the merged clusters with an optimal lattice anonymization process to automatically choose a generalization scheme to provide a k-anonymous dataset with minimal suppression.

14. A computing machine implemented method, comprising:

receiving data at a processor representing a plurality of clusters, wherein each cluster represents a geographic area having a centroid and a population of individuals;

merging a first cluster having a smallest population with a second cluster having a nearest centroid based on a determination that the population of the first cluster is below a minimum size threshold to form an updated cluster that represents an updated geographic area having an updated centroid and an updated population that reflects the merging of the first cluster and the second cluster;

repeating the merging until each cluster meets a minimum size threshold;

de-identifying data records associated with the individuals of the clusters when the smallest cluster meets the minimum size threshold;

assessing a risk of re-identification of the de-identified clusters based on k-anonymity;

increasing the minimum size threshold and repeating the merging, the de-identifying, and the assessing a risk, based on a determination that the assessed risk does not meet a risk criterion; and

presenting the de-identified clusters on a display based on a determination that the assessed risk meets the risk criterion.

15. The computing machine implemented method of claim 14 , wherein the merging comprises:

identifying a smallest one of the clusters based on a number of data records of the cluster; and

merging the smallest cluster with a second one of the clusters.

16. The computing machine implemented method of claim 15 , wherein the merging further comprises:

selecting the second cluster based on a geographic distance between geographic centroids of the smallest cluster and the second cluster.

17. The computing machine implemented method of claim 16 , wherein the selecting comprises:

selecting the second cluster based further on a socio-economic, a demographic, and/or a political parameter.

18. The computing machine implemented method of claim 14 , wherein the assessing comprises:

assessing the risk of re-identification based further on optimal lattice anonymization.

19. The computing machine implemented method of claim 14 , further comprising:

retaining a hierarchy of the merged clusters; and

utilizing the hierarchy of the merged clusters with an optimal lattice anonymization process to automatically choose a generalization scheme to provide a k-anonymous dataset with minimal suppression.