Information criterion-based systems and methods for constructing combining weights for multimodel forecasting and prediction
First Claim
Patent Images
1. A computer-implemented method for generating a weighted average forecast model, comprising:
- receiving, using one or more processors, a plurality of forecasting models;
receiving, using the one or more processors, time series data;
optimizing, using the one or more processors, one or more parameters for each of the received forecasting models, wherein optimizing a parameter includes using the received time series data;
determining, using the one or more processors, an information criteria value for each of the optimized forecasting models, wherein the information criteria value indicates fit quality and complexity;
determining, using the one or more processors, a lowest calculated information criteria value;
determining, using the one or more processors, an information criteria delta value for each of the optimized forecasting models, wherein a delta value indicates a difference between an information criteria value for an optimized forecasting model and the lowest calculated information criteria value;
determining, using the one or more processors, raw weights for each of the optimized forecasting models using the information criteria delta values;
determining, using the one or more processors, normalized weighting factors for each of the optimized forecasting models using the raw weights;
generating, using the one or more processors, a weighted average forecast model using the optimized forecasting models and the normalized weighting factors; and
generating a forecast using the weighted average forecast model.
1 Assignment
0 Petitions
Accused Products
Abstract
Systems and methods are provided for a computer-implemented method for automatically generating a weighted average forecast model that includes receiving a plurality of forecasting models and time series data. At least one parameter of each of the received forecasting models is optimized utilizing the received time series data. A weighting factor is generated for each of the plurality of optimized forecasting models utilizing an information criteria value indicating fit quality of each of the optimized forecasting models, and the generated weighting factors are stored.
37 Citations
39 Claims
-
1. A computer-implemented method for generating a weighted average forecast model, comprising:
-
receiving, using one or more processors, a plurality of forecasting models; receiving, using the one or more processors, time series data; optimizing, using the one or more processors, one or more parameters for each of the received forecasting models, wherein optimizing a parameter includes using the received time series data; determining, using the one or more processors, an information criteria value for each of the optimized forecasting models, wherein the information criteria value indicates fit quality and complexity; determining, using the one or more processors, a lowest calculated information criteria value; determining, using the one or more processors, an information criteria delta value for each of the optimized forecasting models, wherein a delta value indicates a difference between an information criteria value for an optimized forecasting model and the lowest calculated information criteria value; determining, using the one or more processors, raw weights for each of the optimized forecasting models using the information criteria delta values; determining, using the one or more processors, normalized weighting factors for each of the optimized forecasting models using the raw weights; generating, using the one or more processors, a weighted average forecast model using the optimized forecasting models and the normalized weighting factors; and generating a forecast using the weighted average forecast model. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19)
-
-
20. A system for generating a weighted average forecast model, comprising:
-
one or more processors; one or more non-transitory computer-readable storage mediums containing instructions configured to cause the one or more processors to perform operations including; receiving a plurality of forecasting models; receiving time series data; optimizing one or more parameters for each of the received forecasting models, wherein optimizing a parameter includes using the received time series data; determining an information criteria value for each of the optimized forecasting models, wherein the information criteria value indicates fit quality and complexity; determining a lowest calculated information criteria value; determining an information criteria delta value for each of the optimized forecasting models, wherein a delta value indicates a difference between an information criteria value for an optimized forecasting model and the lowest calculated information criteria value; determining raw weights for each of the optimized forecasting models using the information criteria delta values; determining normalized weighting factors for each of the optimized forecasting models using the raw weights; generating a weighted average forecast model using the optimized forecasting models and the normalized weighting factors; and generating a forecast using the weighted average forecast model. - View Dependent Claims (22, 23, 24, 25, 26, 27, 28, 29, 30)
-
-
21. A non-transitory computer program product for generating a weighted average forecast model, tangibly embodied in a machine-readable non-transitory storage medium, including instructions configured to cause a data processing system to:
-
receive a plurality of forecasting models; receive time series data; optimize one or more parameters for each of the received forecasting models, wherein optimizing a parameter includes using the received time series data; determine an information criteria value for each of the optimized forecasting models, wherein the information criteria value indicates fit quality and complexity; determine a lowest calculated information criteria value; determine an information criteria delta value for each of the optimized forecasting models, wherein a delta value indicates a difference between an information criteria value for an optimized forecasting model and the lowest calculated information criteria value; determine raw weights for each of the optimized forecasting models using the information criteria delta values; determine normalized weighting factors for each of the optimized forecasting models using the raw weights; generate a weighted average forecast model using the optimized forecasting models and the normalized weighting factors; and generating a forecast using the weighted average forecast model. - View Dependent Claims (31, 32, 33, 34, 35, 36, 37, 38, 39)
-
Specification