Methods and apparatus for entering and evaluating time dependence hypotheses and for forecasting based on the time dependence hypotheses entered

• US 6,216,154 B1
• Filed: 04/24/1998
• Issued: 04/10/2001
• Est. Priority Date: 04/24/1998
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Ingerman, David, Altschuler, Steven J.
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): Trinh, William
• Application Number:
  US09/065,963
• Time in Patent Office: 1,082 Days
• Field of Search: 707/5, 709/223, 709/217-219, 709/203
• US Class Current: 709/203
• CPC Class Codes: G06F 11/3452 Performance evaluation by statistical analysis
• CPC Class Codes: G06F 2201/875 Monitoring of systems including the internet
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

Entering and evaluating group(s) of similar dependence hypotheses, such as time dependence hypotheses, and forecasting values of a variable based on the group(s) of similar dependence hypotheses. The group(s) of similar dependence hypotheses may be used to forecast requests for an Internet resource or Internet resources having one or more particular attributes. The entered group(s) of similar dependence hypotheses may be evaluated based on known data, such as past requests for Internet resources or Internet resources of a particular type.

Claims

1. In an apparatus having a graphical user interface including a display device and a user input device, a method for entering and displaying at least one group of similar dependency hypotheses related to a temporal relationship between values of first and second variables, wherein the first and second variables represent requests for correspondingly different resources for use in a computer system, the method comprising steps of:

a) displaying, on the display device, a two dimensional rectangle having a first side corresponding to time segments over a period of past time and a second side, perpendicular to the first side, corresponding to time segments over a period of future time;

b) receiving a group of user selections, via the user input device, each of which relates a future time segment on the second side of the rectangle to a past time segment on the first side of the rectangle, thereby forming a group of similar dependency hypotheses such that values of the first variable during each of future time segments will depend on values of the second variable during a related one of each of the past time segments over all of said hypotheses; and

c) displaying, on the display device, points on the two dimensional rectangle corresponding to the group of user selections received, wherein each of the points represents an associated pair of the first and second variables of a corresponding similar dependency hypothesis.

2. The method of claim 1 wherein all points corresponding to the group of selections received are displayed in one color.

3. The method of claim 1 wherein more than one group of selections may be received, and

wherein points corresponding to different groups of selections are displayed, on the display device, in different colors.

4. The method of claim 1 wherein each of the first and second sides of the two dimensional rectangle is divided into time segments over a week.

5. The method of claim 4 wherein each of the time segments is a six-hour time slot.

6. The method of claim 1 further comprising a step of receiving, via the user input device, user definitions of at least one of (a) the period of past time, (b) the time segments of the period of past time, (c) the period of future time, and (d) the time segments of the period of future time.

7. The method of claim 1 further comprising a step of:

d) displaying, on the display device, a bar, wherein a length of the bar corresponds to a correlation of known data corresponding to the group of similar dependency hypotheses.

8. The method of claim 7 wherein all points corresponding to a group of selections received are displayed in one color, and

wherein the bar is displayed in the same color as the points on the two dimensional rectangle corresponding to the group of selections received.

9. The method of claim 1 further comprising a step of:

d) displaying, on the display device, a plot of known data related to the group of similar dependency hypotheses.

10. The method of claim 1 further comprising a step of:

d) displaying, on the display device, a forecast of future values of the variable based on the group of similar dependency hypotheses.

11. The method of claim 1 wherein the values of the variable are numbers of requests for an Internet resource, during particular time periods.

12. The method of claim 1 wherein the values of the variables are numbers of requests for Internet resources of a particular attribute type, during particular time periods.

13. A method for generating a global forecast of a request for a resource for use in a computer system, the method comprising steps of:

a) accepting at least one group of similar dependency hypotheses, wherein each hypothesis of a group assumes that a value of a first variable during a future time period will depend on a value of a second variable during a past time period, wherein the first and second variables represent requests for correspondingly different resources for use in the system;

b) for each one of the at least one group of similar dependency hypotheses accepted;

i) accepting known data related to the one of the at least one group of similar dependency hypotheses;

ii) determining a correlation between the first and second variables of the known data related to the one of the at least one group of similar dependency hypotheses; and

iii) determining a corresponding one weight of the correlation, said one weight being one of a plurality of weights; and

c) determining the global forecast based on each of the at least one group of similar dependency hypotheses, the known data related to each of the at least one group of similar dependency hypotheses, and each of the plurality of weights.

14. The method of claim 13 wherein the weight is determined based on the correlation and a number of known data used to determine the correlation.

15. The method of claim 13 wherein the weight is determined based on the correlation, a number of known data used to determine the correlation, and an average time between the past time periods and the related future time periods of the hypotheses of the one of the at least one group of similar dependency hypotheses.

16. A method for generating a global forecast of a request for a resource for use in a computer system, the method comprising steps of:

a) accepting at least one group of similar dependency hypotheses, wherein each hypothesis of a group assumes that a value of a first variable during a future time period will depend on a value of a second variable during a past time period, wherein the first and second variables represent requests for correspondingly different resources for use in the system;

b) for each one of the at least one group of similar dependency hypotheses accepted;

i) accepting known data related to the one of the at least one group of similar dependency hypotheses;

ii) determining a correlation between the first and second variables of the known data related to the one of the at least one group of similar dependency hypotheses;

iii) determining a corresponding one weight of the correlation, said one weight being one of a plurality of weights; and

iv) forecasting an unknown value based on the one of the at least one group of similar dependency hypotheses and the known data related to the one of the at least one group of similar dependency hypotheses so as to define a corresponding one of a plurality of forecasts; and

c) determining the global forecast based on each of the plurality of said forecasts and each of the plurality of weights.

17. The method of claim 16 wherein the weight is determined based on the correlation and a number of known data used to determine the correlation.

18. The method of claim 16 wherein the weight is determined based on the correlation, a number of known data used to determine the correlation, and an average time between the past time periods and the related future time periods of the hypotheses of the one of the at least one group of similar dependency hypotheses.

19. A method for generating a rank ordered list of groups of similar dependency hypotheses of requests for resources for use in a computer system, the method comprising steps of:

a) accepting groups of similar dependency hypotheses, wherein each hypothesis of a group assumes that a value of a first variable during a future time period will depend on a value of a second variable during a past time period, wherein the first and second variables represent requests for correspondingly different resources for use in the system;

b) for each one of the groups of similar dependency hypotheses accepted;

i) accepting known data related to the one group of the groups of similar dependency hypotheses;

ii) determining a correlation between the first and second variables of known data related to the one group of the groups of similar dependency hypotheses, and

iii) determining a weight of the correlation so as to define a corresponding one of a plurality of weights; and

c) rank ordering the groups of similar dependency hypotheses based on each of the plurality of weights.

20. The method of claim 19 wherein the weight is determined based on the correlation and a number of known data used to determine the correlation.

21. The method of claim 19 wherein the weight is determined based on the correlation, a number of known data used to determine the correlation, and an average time between the past time periods and the related future time periods of the hypotheses of the one group of the groups of similar dependency hypotheses.

22. A device for entering and displaying at least one group of similar dependency hypotheses related to a temporal relationship between values of variables for requests for resources in a computer system, the device comprising:

a) an input device for accepting a user input of the at least one group of similar dependency hypotheses, wherein each hypothesis of a group assumes that a value of a first variable during a future time period will depend on a value of a second variable during a past time period, wherein the first and second variables represent requests for correspondingly different resources for use in the system;

b) a hypothesis generator, coupled with the input device, for processing the user input of the at least one group of similar dependency hypotheses;

c) an output device, coupled with the hypothesis generator, for rendering the user input of the at least one group of similar dependency hypotheses; and

d) a hypothesis correlator, coupled with the input device, for receiving known data and for determining a correlation between the first and second variables of the known data related to each of the at least one group of similar dependency hypotheses.

23. The device of claim 22 wherein the output device further renders the at least one correlation determined.

24. The device of claim 22 further comprising:

e) a weight generator, coupled with the hypothesis correlator, for determining a weight corresponding each of the at least one correlation.

25. The device of claim 24 wherein the weight is determined based on the correlation, a number of known data used to determine the correlation, and an average time between the past time periods and the related future time periods of the hypotheses of the one of the at least one group of similar dependency hypotheses.

26. The device of claim 22 further comprising:

e) a weight generator, coupled with the hypothesis correlator, for determining a weight corresponding each of the at least one correlation, based on the correlation and a size of the known data used to determine the correlation.

27. The device of claim 26 further comprising:

f) a forecast generator, coupled with (i) the hypothesis correlator and (ii) the weight generator, for forecasting a value based on the known data, the at least one weight, and the at least one correlation.

28. The device of claim 26 further comprising:

f) a forecast generator, coupled with (i) the hypothesis correlator and (ii) the weight generator, for forecasting a value based on the known data, each of the at least one weight, and each of the at least one correlation.