Determining confidence intervals for weighted trial data

• US 7,092,844 B1
• Filed: 07/20/2004
• Issued: 08/15/2006
• Est. Priority Date: 07/20/2004
• Status: Active Grant

• Current Assignee: Versata Development Group, Inc. (ESW Capital, LLC)
• Original Assignee: Trilogy Development Group Incorporated (ESW Capital, LLC)
• Inventors: Young, Stephen W., Karipides, Daniel P.
• Primary Examiner(s): Raymond, Edward
• Application Number:
  US10/896,162
• Time in Patent Office: 756 Days
• Field of Search: 702/176, 702/179, 702/181
• US Class Current: 702/181
• 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

Abstract

Collected trial data is weighted to, for example, reflect the relevance of the data. A weighted confidence interval determination and application process described herein determines a confidence interval used to interpret the data. The confidence interval is derived by adjusting the sample size N to account for the impact of weighting when determining confidence intervals. The sample size N is adjusted in a downward trend to avoid overestimating the confidence interval. Lower and upper bounds of the confidence interval are determined using weight influenced variables. Thus, interpretation of weighted trial data, such as product demand data, can be achieved and acted upon with an accurate estimation of risk.

Claims

1. A method for using a computer system to reduce overestimation of confidence intervals determined from weighted trial data, the method comprising:

adjusting a total number of trials N in downward trend using a factor derived from trial data weights;

determining a confidence interval using the adjusted total number of trials;

interpreting the weighted trial data; and

assigning the confidence interval to the interpreted data.

2. The method of claim 1 wherein the factor derived from trial data weights comprises a normalizing factor derived from weights of each trial set of data.

3. The method of claim 2 wherein the normalizing factor comprises:

4. The method of claim 1 wherein the confidence interval is bounded by respective upper and lower bounds.

5. The method of claim 4 wherein the upper and lower bounds of the confidence interval for an outcome i are derived from an expected probability Θ

_i of outcome i occurring in any random trial, and expected probability Θ

_i is derived from trial data weights from each of the N trials.

6. The method of claim 5 further comprising determining the probability Θ

_i in accordance with;

7. The method of claim 1 wherein the interpreted data comprises data representing demand data for one or more products.

8. The method of claim 7 further comprising:

generating product demand data reports that include the interpreted data and assigned confidence interval.

9. The method of claim 7 wherein the one or more products are vehicles.

10. The method of claim 1 further comprising:

displaying the interpreted data including the assigned confidence interval.

11. The method of claim 1 further comprising:

collecting trial data from internet session data and applying weights to samples of the trial data to generated the weighted trial data.

12. The method of claim 1 wherein the trial data weights from which the factor is derived are non-zero.

13. A computer system for reducing overestimation of confidence intervals determined from weighted trial data, the computer system comprising:

a processor; and

a computer program storage medium coupled to the processor and having code encoded therein, wherein the code is executable by the processor and comprises code to;

adjust a total number of trials N in downward trend using a factor derived from trial data weights;

determine a confidence interval using the adjusted total number of trials;

interpret the weighted trial data; and

assign the confidence interval to the interpreted data.

14. The computer system of claim 13 wherein the factor derived from trial data weights comprises a normalizing factor derived from weights of each trial set of data.

15. The computer system of claim 14 wherein the normalizing factor comprises:

16. The computer system of claim 13 wherein the confidence interval is bounded by respective upper and lower bounds.

17. The computer system of claim 16 wherein the upper and lower bounds of the confidence interval for an outcome i are derived from an expected probability Θ

_i of outcome i occurring in any random trial, and expected probability Θ

_i is derived from trial data weights from each of the N trials.

18. The computer system of claim 17 further comprising code to determine the probability Θ

_i in accordance with;

19. The computer system of claim 13 wherein the interpreted data comprises data representing demand data for one or more products.

20. The computer system of claim 19 further comprising code to generate product demand data reports that include the interpreted data and assigned confidence interval.

21. The computer system of claim 19 wherein the one or more products are vehicles.

22. The computer system of claim 13 further comprising code to cause a display of the interpreted data including the assigned confidence interval.

23. The computer system of claim 13 further comprising code to perform an automated collection of internet session trial data and applying weights to samples of the trial data to generated the weighted trial data.

24. The computer system of claim 13 wherein the trial data weights from which the factor is derived are non-zero.

25. A computer system for interpreting weighted trial data using a non-overestimated confidence interval, the computer system comprising:

means for adjusting a total number of trials N in downward trend using a factor derived from trial data weights;

means for determining the confidence interval using the adjusted total number of trials;

means for interpreting the weighted trial data; and

means for assigning the confidence interval to the interpreted data.

26. The computer system of claim 25 wherein the factor derived from trial data weights comprises a normalizing factor derived from weights of each trial set of data and the normalizing factor comprises:

27. The computer system of claim 25 wherein the confidence interval is bounded by respective upper and lower bounds and the upper and lower bounds of the confidence interval for an outcome i are derived from an expected probability Θ

_i of outcome i occurring in any random trial, and expected probability Θ

_i is derived from trial data weights from each of the N trials.

28. The computer system of claim 25 wherein the trial data weights from which the factor is derived are non-zero.