APPARATUS AND METHOD FOR PROBABILISTIC POPULATION SIZE AND OVERLAP DETERMINATION, REMOTE PROCESSING OF PRIVATE DATA AND PROBABILISTIC POPULATION SIZE AND OVERLAP DETERMINATION FOR THREE OR MORE DATA SETS

• US 20060241903A1
• Filed: 06/20/2006
• Published: 10/26/2006
• Est. Priority Date: 02/04/1997
• Status: Abandoned Application

• Current Assignee: The Bristol Observatory Limited
• Original Assignee: The Bristol Observatory Limited
• Inventors: Pandiani, John A., Banks, Steven M
• Application Number:
  US11/425,180
• Publication Number:
  US 20060241903A1
• Time in Patent Office: Days
• US Class Current: 702/179
• CPC Class Codes: G06F 16/2462 Approximate or statistical queries
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G16H 10/20 for electronic clinical trials or questionnaires
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching

Abstract

The invention determines the population size and population overlap in data containing records on the unique entities without unique identifiers for the unique entities and having at least one common type of information with a known distribution of finite expectation by decomposing probabilistic calculations. The computer determines population overlap of unique entities between the data sets by subtracting a probabilistic incremental number of unique entities needed for a larger total number of values of the information with the known distribution from the data sets. The invention can also maintain the security of private data by allowing a remote computer where the original data is stored to download diagnostic and aggregation procedures from another computer over a network. The remote computer performs the functions on the data and forwards the results to the estimate processor computer over the network. The estimate processor determines population size and overlap from aggregate results and forwards this information back to the remote computer over the network. The invention also determines the overlap of three or more data sets by concatenating all combinations of the data sets and determining estimates for all subsets of the combinations of the data sets. The operations involve the cancellation of equivalent terms that have opposite signs.

Claims

1. A method of determining population size of unique entities in data comprising aggregate data of two different data sets that overlap and containing records on unique entities without unique identifiers for the unique entities, comprising:

receiving the aggregate data comprising data set subset information and distribution occurrence information from a holder of the data; and

determining a population size estimate from the aggregate data.

2. A method as recited in claim 1, wherein the unique entities comprise human individuals.

3. A method as recited in claims 2 where the aggregate data comprises a number of dates of birth for each year of birth and gender cohort in each data set.

4. A method as recited in claim 1, further comprising transferring the population size to the holder.

5. A method as recited in claim 1, further comprising

determining a population overlap estimate of data sets in the data from the received aggregate data;

and

transferring the overlap to the holder.

6. A method as recited in claim 1, further comprising:

diagnosing, by the holder, data record errors in the data; and

creating, by the holder, a combined data set from data sets of the data and determining the aggregate data from the data sets and the combined data set.

7. A method as recited in claim 6, further comprising:

transferring a diagnostic procedure for performing the diagnosing and an aggregate procedure for performing the creating to the holder.

8. A method as recited in claim 1, wherein the diagnostic procedure deletes invalid records.

9. A distributed method of determining population size in overlapping data sets, comprising:

performing data diagnostics and aggregation of data in the data sets at a first location; and

;

determining population size at a second location different from the first location.

10. A computer readable storage controlling a computer via a process of receiving aggregate data comprising data set subset and distribution occurrence information from a holder of the data and determining a population size estimate from the aggregate data.

11. A computer readable storage controlling a computer via a data structure comprising a list of records comprising for a record containing subset identifier fields and records containing subset labels and a number of occurrences fields.

12. A method of diagnosing errors in data sets containing records on unique entities without unique identifiers for the unique entities and having at least one common type of information with a known distribution of finite expectation, comprising:

determining whether the data has valid values;

determining whether the records indicate a variance from the known distribution; and

determining whether the data fills a maximum number of subdivisions of the data in the data sets.

13. A method as recited in claim 12, further comprising removing records containing errors.

14. A method of aggregating data of data sets containing records on unique entities without unique identifiers for the unique entities and having at least one common type of information with a known distribution of finite expectation, comprising:

creating a combined data set comprising all records from the previous data set; and

producing aggregate data comprising data set subset and distribution occurrence information.

15. A method of determining a population size of unique entities in three or more data sets containing records on the unique entities without unique identifiers for the unique entities, said method comprising:

producing all of the N-way concatenations of the data sets where N equals from 2 to a total number of data sets; and

performing population size estimate calculations using the concatenations of the data sets.

16. A method as recited in claim 15, wherein the performing comprises canceling equivalent terms of opposite sign.

17. A method as recited in claim 15, further comprising performing population size overlap estimate calculations.

18. A method as recited in claim 15, wherein the performing comprises:

estimating unions of concatenated data sets; and

performing a probabilistic decomposition of the unions.

19. A method of determining a population size of unique entities in three or more data sets containing records on the unique entities without unique identifiers for the unique entities, said method comprising:

producing all of the N-way concatenations of the data sets where N equals from 2 to a total number of data sets; and

performing population overlap and size estimate calculations using the concatenations of the data sets comprising canceling equivalent terms of opposite sign for all of the sets of the data sets.

20. A computer readable storage controlling a computer via a process of producing all of the N-way concatenations of the data sets where N equals from 2 to a total number of data sets and performing estimate calculations using the concatenations of the data sets.

21. An apparatus for population size and overlap determination of three or more data sets, comprising a computer producing all of the N-way concatenations of the data sets where N equals from 2 to a total number of data sets and performing population size and overlap estimate calculations using the concatenations of the data sets for all subsets of the data sets.

22. An apparatus for probabilistic population size determination, comprising a computer probabilistically calculating the population size of unique entities in data, containing records on unique entities without unique identifiers for the unique entities and having at least one common type of information with a known distribution of finite expectation, using decomposed probabilistic calculations based on values of the information with the known distribution.

23. A method using a computer to probabilistically determine a population size of unique entities in data containing records on the unique entities without unique identifiers for the unique entities and having at least one common type of information with a known distribution of finite expectation, said method comprising decomposing probabilistic calculations based on values of the information with the known distribution.

24. A computer program embodied on a computer-readable medium for probabilistically calculating a population size and a population overlap of unique entities in first and second data sets containing records on the unique entities without unique identifiers for the unique entities and having at least one common type of information with a known distribution of finite expectation, said computer program comprising:

a data preparation segment combining the first and second data sets into a combined data set;

a population size measurement segment probabilistically calculating the population size for the first and second data sets using decomposed probabilistic calculations based on values of the information with the known distribution; and

, a population overlap measurement segment determining the population overlap of the unique entities between the first and second data sets by subtracting a probabilistic incremental number of unique entities needed for a larger total number of values of the information with the known distribution from either of the first and second data sets to increase to a total number of values of the information with the known distribution in the combined data set from a smaller of the population size of the first and second data sets.

25. An apparatus for population size determination, comprising a computer calculating the population size of unique entities in a database, containing records on the unique entities without unique identifiers for the unique entities and having at least one common type of information with a known distribution of finite expectation, using decomposed probabilistic calculations based on values of the information with the known distribution.