Computer-implemented systems and methods for defining events

• US 8,005,707 B1
• Filed: 05/09/2006
• Issued: 08/23/2011
• Est. Priority Date: 05/09/2005
• Status: Active Grant

• Current Assignee: SAS Institute Incorporated
• Original Assignee: SAS Institute Incorporated
• Inventors: Jackson, Wilma S., Elsheimer, David Bruce, Leonard, Michael J.
• Primary Examiner(s): Boswell; Beth V
• Assistant Examiner(s): Dickerson; Tiphany B
• Application Number:
  US11/431,127
• Time in Patent Office: 1,932 Days
• Field of Search: 705/29, 705/31, 702179-181, 702/189, 703/2, 703/22, 700 29- 31
• US Class Current: 705/7.29
• 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 10/10 Office automation; Time management
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities

Abstract

Computer-implemented systems and methods to process time-series data. As an example, a system and method can use a first data store to store time series data, and a second data store to store definitions of events. A dummy variable is generated when data from an event'"'"'s definition is applied to time series data.

Claims

1. A computer-implemented method for processing time-series data, comprising:

accessing, using one or more data processors, time series data including one or more time series;

accessing, using the one or more data processors, one or more event definitions, wherein each event definition corresponds to an association between an event and a set of one or more parameters;

generating, using the one or more data processors, one or more dummy variables for a time series, wherein a dummy variable is generated by applying an event to the time series, and wherein an event definition corresponding to the event is used to apply the event to the time series;

identifying, using the one or more data processors, a model including the event;

forecasting, using the one or more data processors, the time series, wherein forecasting includes fitting the model to the time series using the event definition corresponding to the event and the one or more dummy variables;

adjusting, using the one or more data processors, the one or more dummy variables when the time series is altered, wherein adjusting includes applying the event to the altered time series to automatically generate one or more adjusted dummy variables; and

forecasting, using the one or more data processors, the altered time series, wherein forecasting includes fitting the identified model to the altered time series using the event definition corresponding to the event and the one or more adjusted dummy variables.

2. The method of claim 1, wherein the time series data includes an initial time, an ending time, and a frequency.

3. The method of claim 1, further comprising:

optimizing one of the one or more parameters associated with the event.

4. The method of claim 2 further comprising:

selecting the event or an event group based upon the frequency of the time series data.

5. The method of claim 4, wherein the frequency is a seasonal frequency.

6. The method of claim 2, wherein the altered time series includes a different starting time and a different ending time than the time series.

7. The method of claim 1, further comprising:

adjusting the one or more dummy variables when the event is altered.

8. The method of claim 1, wherein the one or more dummy variables indicate how the event impacts the forecasted time series.

9. The method of claim 1, wherein the set of one or more parameters includes event name parameters, class parameters, key name parameters, timing value parameters, event type parameters, value parameters, shift parameters, pulse parameters, duration parameters, slope parameters, period parameters, controls in an exponential curve, or event combination rules.

10. The method of claim 1, wherein the event is an Easter event, a United States Thanksgiving event, a boxing event, a Christmas event, a father'"'"'s day event, a Halloween event, a United States Independence day event, a labor day event, a memorial day event, a mother'"'"'s day event, a new year'"'"'s day event, a Canadian Thanksgiving day event, or a valentine'"'"'s day event.

11. The method of claim 1, wherein the time series data is stored in a first data store, and wherein the one or more event definitions are stored in a second separate data store.

12. The method of claim 1, wherein the time series includes an ID statement and one or more time ID values, and wherein the event is also based upon the ID statement and the one or more time ID values.

13. The method of claim 12, wherein the one or more time ID values are defined using a formula.

14. The method of claim 1, wherein the identified model is stored for subsequent use.

15. A system for processing time-series data, comprising:

one or more processors;

one or more computer readable storage mediums containing instructions configured to cause the one or more processors to perform operations including;

accessing time series data including one or more time series;

accessing one or more event definitions, wherein each event definition corresponds to an association between an event and a set of one or more parameters;

generating one or more dummy variables for a time series, wherein a dummy variable is generated by applying an event to the time series, and wherein an event definition corresponding to the event is used to apply the event to the time series;

identifying a model including the event;

forecasting the time series, wherein forecasting includes fitting the model to the time series using the event definition corresponding to the event and the one or more dummy variables;

adjusting the one or more dummy variables in response to an alteration of the time series, wherein adjusting includes applying the event to the altered time series to automatically generate one or more adjusted dummy variables; and

forecasting the altered time series, wherein forecasting includes fitting the identified model to the altered time series using the event definition corresponding to the event and the one or more adjusted dummy variables.

16. A computer program product for processing time-series data, tangibly embodied in a machine readable storage medium, including instructions configured to cause a data processing apparatus to:

access time series data including one or more time series;

access one or more event definitions, wherein each event definition corresponds to an association between an event and a set of one or more parameters;

generate one or more dummy variables for a time series, wherein a dummy variable is generated by applying an event to the time series, and wherein an event definition corresponding to the event is used to apply the event to the time series;

identify a model including the event;

forecast the time series, wherein forecasting includes fitting the model to the time series using the event definition corresponding to the event and the one or more dummy variables;

adjust the one or more dummy variables in response to an alteration of the time series, wherein adjusting includes applying the event to the altered time series to automatically generate one or more adjusted dummy variables; and

forecast the altered time series, wherein forecasting includes fitting the identified model to the altered time series using the event definition corresponding to the event and the one or more adjusted dummy variables.