Methods and systems for forecasting with model-based PDF estimates

• US 8,180,664 B2
• Filed: 12/01/2004
• Issued: 05/15/2012
• Est. Priority Date: 12/01/2004
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Shan, Jerry Z.
• Primary Examiner(s): Boswell, Beth V
• Assistant Examiner(s): DICKERSON, TIPHANY B
• Application Number:
  US11/001,453
• Publication Number:
  US 20060116920A1
• Time in Patent Office: 2,722 Days
• Field of Search: 705/7.31
• 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 10/063 Operations research, analysis or management
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities

Abstract

Disclosed herein are systems and methods for forecasting with model-based PDF (probability density function) estimates. Some method embodiments may comprise: estimating model parameters for a time series, calculating a PDF for the time series, and generating a forecast from the PDF. The model parameters may comprise a variance for a hidden noise source, and the PDF for the time series may be based at least in part on an estimated variance for the hidden noise source.

Claims

1. A computer-readable storage medium storing a program that, when executed by a processor, causes the processor to:

select a reference set of profiles from previous periods;

estimate model parameters of a time series based on the reference set, wherein the model parameters comprise a first variance for a hidden noise source;

calculate a probability density function for the time series including determining a second variance for the probability density function based at least in part on the first variance for the hidden noise source; and

generate a forecast from the probability density function.

2. The computer-readable storage medium of claim 1, wherein when the processor estimates, the program further causes the processor to differentiate the time series to obtain a stationary time series.

3. The computer-readable storage medium claim 1, wherein when the processor estimates, the program further causes the processor to estimate using model parameters comprising coefficients of a moving average filter.

4. The computer-readable storage medium of claim 1, wherein when the processor estimates, the program further causes the processor to estimate using model parameters comprising coefficients of an autoregressive filter.

5. The computer-readable storage medium of claim 1, wherein when the processor estimates, the program further causes the processor to estimate using model parameters comprising an offset value.

6. The computer-readable storage medium of claim 1, wherein when the processor estimates, the program further causes the processor to estimate using model parameters where the model accounts for recurring patterns in the time series.

7. The computer-readable storage medium of claim 1, wherein when the processor calculates, the program further causes the processor to calculate a Gaussian probability density function.

8. The computer-readable storage medium of claim 1, wherein when the processor calculates, the program further causes the processor to calculate the probability density function being a function of time.

9. The computer-readable storage medium of claim 1, further comprising:

wherein when the processor estimates the time series, the program further causes the processor to estimate the time series representing a cumulative value; and

wherein when the processor generates the forecast, the processor generates the forecast being an end-of-period cumulative value.

11. The computer-readable storage medium of claim 1 wherein when the processor changes the profiles, the program further causes the processor to perform a dissimilarity calculations between the profile of the current period, and to change the profiles in the reference set based on the dissimilarity calculations.

10. A computer comprising:

a display;

a processor coupled to the display; and

a memory coupled to the processor,

wherein the memory stores software that configures the processor to;

select reference profiles from a set of profiles from previous periods;

estimate a time series based on the reference profiles and a profile of the current period; and

derive a probability density function for the time series by estimating parameters of a model that comprises a hidden noise source,

wherein the software configures the processor to determine a first variance for the hidden noise source, and wherein the software further configures the processor to determine a second variance for the probability density function from the first variance of the hidden noise source and from estimated filter coefficients.