System and method for adaptive forecasting

• US 7,899,699 B1
• Filed: 11/04/2004
• Issued: 03/01/2011
• Est. Priority Date: 11/04/2004
• Status: Active Grant

• Current Assignee: American Airlines Incorporated (American Airlines Group Incorporated)
• Original Assignee: American Airlines Incorporated (American Airlines Group Incorporated)
• Inventors: Kumar, Praveen
• Primary Examiner(s): Jasmin; Lynda C
• Assistant Examiner(s): MILLER, ALAN S
• Application Number:
  US10/981,902
• Time in Patent Office: 2,308 Days
• Field of Search: 705 7- 10
• US Class Current: 705/7.31
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities
• CPC Class Codes: G06Q 50/14 Travel agencies

Abstract

A system and method for adaptive forecasting is provided. A model file containing a forecast model is generated. A calibration type that correspond to a symbol and a key in the forecast model is determined, and the model file is stored in a database. A request for a forecast based on the symbol and the key is received. The model file is retrieved from the database and a forecast is calculated using the forecast model.

Claims

1. A method for adaptive forecasting, comprising:

receiving a request to forecast a number of tickets that will be sold for a future airline flight, wherein the future airline flight will depart;

from a departure market,

at a departure time,

on a day of the week, and

in a month in a quarter;

retrieving a first permanent component, the first permanent component being a historical average number of tickets that were sold over a first time period for past airline flights that have departed from the departure market at the departure time;

retrieving a first trend component, the first trend component being based on a first trend difference between the first permanent component and a first previous permanent component,

the first previous permanent component being a historical average number of tickets that were sold over a second time period for past airline flights that have departed from the departure market at the departure time,

wherein the end of the second time period occurred before the end of the first time period,

wherein the first trend difference is positive if the first permanent component is greater than the first previous permanent component, and

wherein the first trend difference is negative if the first permanent component is less than the first previous permanent component;

retrieving a first seasonal component,

the first seasonal component being based on a first seasonal difference between the first permanent component and a historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at the departure time, and that have departed on the day of the week, in the month, or in the quarter,

wherein the first seasonal difference is negative if the first permanent component is greater than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at the departure time, and that have departed on the day of the week, in the month, or in the quarter, and

wherein the first seasonal difference is positive if the first permanent component is less than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at the departure time, and that have departed on the day of the week, in the month, or in the quarter;

determining, using a computer, a first forecast of the number of tickets that will be sold for the future airline flight, the first forecast being based on the first permanent component, the first trend component, and the first seasonal component;

determining whether the first forecast is valid;

and

if the first forecast is not valid, then;

retrieving a second permanent component, the second permanent component being a historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at either the departure time or one or more other departure times;

retrieving a second trend component, the second trend component being based on a second trend difference between the second permanent component and a second previous permanent component,

the second previous permanent component being a historical average number of tickets that were sold over the second time period for past airline flights that have departed from the departure market at either the departure time or the one or more other departure times,

wherein the second trend difference is positive if the second permanent component is greater than the second previous permanent component, and

wherein the second trend difference is negative if the second permanent component is less than the second previous permanent component;

retrieving a second seasonal component,

the second seasonal component being based on a second seasonal difference between the second permanent component and a historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter;

wherein the second seasonal difference is negative if the second permanent component is greater than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter, and

wherein the second seasonal difference is positive if the second permanent component is less than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter; and

determining, using a computer, a second forecast of the number of tickets that will be sold for the future airline flight, the second forecast being based on the second permanent component, the second trend component, and the second seasonal component.

2. The method of claim 1, further comprising:

if the first forecast is not valid, then returning the second forecast in response to the request to forecast the number of tickets that will be sold for the future airline flight.

3. The method of claim 1, further comprising:

if the first forecast is not valid, then;

determining whether the second forecast is valid; and

if the second forecast is valid, then returning the second forecast in response to the request to forecast the number of tickets that will be sold for the future airline flight.

4. The method of claim 1, further comprising:

if the first forecast is valid, then returning the first forecast in response to the request to forecast the number of tickets that will be sold for the future airline flight.

5. The method of claim 1, further comprising:

if the first forecast is valid, then;

determining a third forecast, wherein the third forecast is the product of the first forecast multiplied by a ratio; and

returning the third forecast in response to the request to forecast the number of tickets that will be sold for the future airline flight.

6. The method of claim 5, wherein determining the third forecast comprises:

retrieving a third permanent component, the third permanent component being a historical average number of tickets that were sold over the first time period for past airline flights that have departed from either the departure market or one or more other markets at either the departure time or the one or more other departure times;

retrieving a third trend component, the third trend component being based on a third trend difference between the third permanent component and a third previous permanent component,

the third previous permanent component being a historical average number of tickets that were sold over the second time period for past airline flights that have departed from either the departure market or the one or more other markets at either the departure time or the one or more other departure times,

wherein the third trend difference is positive if the third permanent component is greater than the third previous permanent component, and

wherein the third trend difference is negative if the third permanent component is less than the third previous permanent component;

retrieving a third seasonal component,

the third seasonal component being based on a third seasonal difference between the third permanent component and a historical average number of tickets that were sold over the first time period for past airline flights that have departed from either the departure market or the one or more other markets at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter, and

wherein the third seasonal difference is negative if the third permanent component is greater than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from either the departure market or the one or more other markets at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter, and

wherein the third seasonal difference is positive if the third permanent component is less than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from either the departure market or the one or more other markets at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter;

determining, using a computer, a fourth forecast of the number of tickets that will be sold for the future airline flight, the third forecast being based on the third permanent component, the third trend component, and the third seasonal component; and

determining, using a computer, the ratio, wherein the ratio equals the quotient of the first forecast divided by the fourth forecast.

7. A computer readable medium that is non-transitory, the computer readable medium comprising a plurality of instructions for execution by at least one computer processor, the plurality of instructions comprising:

instructions for receiving a request to forecast a number of tickets that will be sold for a future airline flight, wherein the future airline flight will depart;

from a departure market,

at a departure time,

on a day of the week, and

in a month in a quarter;

instructions for retrieving a first permanent component, the first permanent component being a historical average number of tickets that were sold over a first time period for past airline flights that have departed from the departure market at the departure time;

instructions for retrieving a first trend component, the first trend component being based on a first trend difference between the first permanent component and a first previous permanent component,

the first previous permanent component being a historical average number of tickets that were sold over a second time period for past airline flights that have departed from the departure market at the departure time,

wherein the end of the second time period occurred before the end of the first time period,

wherein the first trend difference is positive if the first permanent component is greater than the first previous permanent component, and

wherein the first trend difference is negative if the first permanent component is less than the first previous permanent component;

instructions for retrieving a first seasonal component,

the first seasonal component being based on a first seasonal difference between the first permanent component and a historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at the departure time, and that have departed on the day of the week, in the month, or in the quarter,

wherein the first seasonal difference is negative if the first permanent component is greater than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at the departure time, and that have departed on the day of the week, in the month, or in the quarter, and

wherein the first seasonal difference is positive if the first permanent component is less than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at the departure time, and that have departed on the day of the week, in the month, or in the quarter;

instructions for determining a first forecast of the number of tickets that will be sold for the future airline flight, the first forecast being based on the first permanent component, the first trend component, and the first seasonal component;

instructions for determining whether the first forecast is valid; and

instructions for if the first forecast is not valid, then;

retrieving a second permanent component, the second permanent component being a historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at either the departure time or one or more other departure times;

retrieving a second trend component, the second trend component being based on a second trend difference between the second permanent component and a second previous permanent component,

the second previous permanent component being a historical average number of tickets that were sold over the second time period for past airline flights that have departed from the departure market at either the departure time or the one or more other departure times,

wherein the second trend difference is positive if the second permanent component is greater than the second previous permanent component, and

wherein the second trend difference is negative if the second permanent component is less than the second previous permanent component;

retrieving a second seasonal component,

the second seasonal component being based on a second seasonal difference between the second permanent component and a historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter;

wherein the second seasonal difference is negative if the second permanent component is greater than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter, and

wherein the second seasonal difference is positive if the second permanent component is less than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from the departure market at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter; and

determining, using a computer, a second forecast of the number of tickets that will be sold for the future airline flight, the second forecast being based on the second permanent component, the second trend component, and the second seasonal component.

8. The computer readable medium of claim 7, wherein the plurality of instructions further comprises:

instructions for if the first forecast is not valid, then returning the second forecast in response to the request to forecast the number of tickets that will be sold for the future airline flight.

9. The computer readable medium of claim 7, wherein the plurality of instructions further comprises:

instructions for if the first forecast is not valid, then;

determining whether the second forecast is valid; and

if the second forecast is valid, then returning the second forecast in response to the request to forecast the number of tickets that will be sold for the future airline flight.

10. The computer readable medium of claim 7, wherein the plurality of instructions further comprises:

instructions for if the first forecast is valid, then returning the first forecast in response to the request to forecast the number of tickets that will be sold for the future airline flight.

11. The computer readable medium of claim 7, wherein the plurality of instructions further comprises:

instructions for if the first forecast is valid, then;

determining a third forecast, wherein the third forecast is the product of the first forecast multiplied by a ratio; and

returning the third forecast in response to the request to forecast the number of tickets that will be sold for the future airline flight.

12. The computer readable medium of claim 11, wherein instructions for determining the third forecast comprise:

instructions for retrieving a third permanent component, the third permanent component being a historical average number of tickets that were sold over the first time period for past airline flights that have departed from either the departure market or one or more other markets at either the departure time or the one or more other departure times;

instructions for retrieving a third trend component, the third trend component being based on a third trend difference between the third permanent component and a third previous permanent component,

the third previous permanent component being a historical average number of tickets that were sold over the second time period for past airline flights that have departed from either the departure market or the one or more other markets at either the departure time or the one or more other departure times,

wherein the third trend difference is positive if the third permanent component is greater than the third previous permanent component, and

wherein the third trend difference is negative if the third permanent component is less than the third previous permanent component;

instructions for retrieving a third seasonal component,

the third seasonal component being based on a third seasonal difference between the third permanent component and a historical average number of tickets that were sold over the first time period for past airline flights that have departed from either the departure market or the one or more other markets at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter, and

wherein the third seasonal difference is negative if the third permanent component is greater than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from either the departure market or the one or more other markets at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter, and

wherein the third seasonal difference is positive if the third permanent component is less than the historical average number of tickets that were sold over the first time period for past airline flights that have departed from either the departure market or the one or more other markets at either the departure time or the one or more other departure times, and that have departed on the day of the week, in the month, or in the quarter;

instructions for determining a fourth forecast of the number of tickets that will be sold for the future airline flight, the third forecast being based on the third permanent component, the third trend component, and the third seasonal component; and

instructions for determining the ratio, wherein the ratio equals the quotient of the first forecast divided by the fourth forecast.