# US 9,519,912 B2 · Hybrid system for demand prediction

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 9,519,912 B2
- Document ID: US9519912B2
- Status: Granted
- Kind: B2
- Application number: 14/032,476
- Filed: 2013-09-20
- Published: 2015-03-26
- Granted: 2016-12-13
- Earliest claimed: 2013-09-20
- Assignees: Xerox Corporation
- Applicants: Xerox Corporation
- Inventors: Boris Chidlovskii; Mohammad Ghufran
- Web page: https://patentgrep.com/patent/US9519912B2

## CPC

- G06Q 30/0202: Market predictions or forecasting for commercial activities
- G06N 20/00: Machine learning
- G06N 20/10: using kernel methods, e.g. support vector machines [SVM]
- G06Q 10/0631: Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

In demand prediction, a history of demand for a resource is modeled to generate a baseline model of the demand, and demand for the resource at a prediction time is predicted by evaluating a regression function of depth k operating on an input data set including at least the demand for the resource at the prediction time output by the baseline model and measured demand for the resource measured at k times prior to the prediction time. The resource may be off-street parking, and the input data set may further include weather data. The regression function may comprise a support vector regression (SVR) function that is trained on the history of demand for the resource. The baseline model suitably comprises a Fourier model of the history of demand for the resource.

## Claims

### Claim 1

A non-transitory storage medium storing instructions readable and executable by an electronic data processing device to perform a method comprising:
generating a Fourier model of history of demand for a resource comprising time series data;
estimating demand for the resource at a prediction time by evaluating the Fourier model at the prediction time; and
generating a prediction of demand for the resource at the prediction time by evaluating a regression function operating on an input data set including at least the estimated demand for the resource at the prediction time estimated by evaluating the Fourier model at the prediction time and measured values of demand for the resource measured at one or more times prior to the prediction time.

### Claim 2 (depends on 1)

The non-transitory storage medium of claim 1 wherein the resource is one of: off-street parking, electrical power, gasoline, and a retail merchandise product.

### Claim 3 (depends on 1)

The non-transitory storage medium of claim 1 wherein the resource is off-street parking and the input data set further includes weather data measured at the one or more times prior to the prediction time.

### Claim 4 (depends on 1)

The non-transitory storage medium of claim 1 wherein the regression function comprises a support vector regression (SVR) function trained on the history of the demand.

### Claim 5 (depends on 1)

The non-transitory storage medium of claim 1 wherein the method further comprises:
training the regression function on the history of the demand.

### Claim 6 (depends on 1)

The non-transitory storage medium of claim 1 wherein the generating of the Fourier model comprises computing Fourier components for a plurality of different periods including at least two of (1) one day, (2) one week, and (3) one year.

### Claim 7 (depends on 1)

The non-transitory storage medium of claim 1 wherein the generating of the Fourier model comprises computing a Fourier transform of the history of the demand.

### Claim 8

An apparatus comprising:
an electronic data processing device configured to receive occupancy data from an off-street parking resource, the electronic data processing device further configured to:
estimate demand for the off-street parking resource at a prediction time based on a Fourier model of historical occupancy data received from the off-street parking resource; and
generate a predicted demand for the off-street parking resource at a prediction time by evaluating a regression function operating on an input data set including at least the estimated demand for the off-street parking resource at the prediction time estimated based on the Fourier model and occupancy data received from the off-street parking resource at k times prior to the prediction time.

### Claim 9 (depends on 8)

The apparatus of claim 8 wherein the input data set further includes weather data measured at the k times prior to the prediction time.

### Claim 10 (depends on 8)

The apparatus of claim 8 wherein the regression function comprises a support vector regression (SVR) function trained on the historical occupancy data.

### Claim 11 (depends on 8)

The apparatus of claim 8 wherein the electronic data processing device is further configured to generate the Fourier model of the historical occupancy data.

### Claim 12 (depends on 11)

The apparatus of claim 11 wherein the electronic data processing device is configured to generate the Fourier model of the historical occupancy data by operations including computing a Fourier component for a time interval of one day and computing a Fourier component for a time interval of one week.

### Claim 13 (depends on 11)

The apparatus of claim 11 wherein the electronic data processing device is configured to generate the Fourier model of the historical occupancy data by computing a Fourier transform of the historical occupancy data.

### Claim 14 (depends on 8)

The apparatus of claim 8 wherein the electronic data processing device is further configured to train the regression function on the historical occupancy data.

### Claim 15

A method comprising:
modeling a history of demand for a resource to generate a baseline model of the demand;
predicting demand for the resource at a prediction time by evaluating a regression function of depth k operating on an input data set including at least the demand for the resource at the prediction time output by the baseline model of the demand and measured demand for the resource measured at k times prior to the prediction time, where k is a positive integer greater than or equal to one;
wherein the modeling and the predicting are performed by an electronic data processing device.

### Claim 16 (depends on 15)

The method of claim 15 wherein k is greater than or equal to two.

### Claim 17 (depends on 15)

The method of claim 15 wherein the resource is off-street parking and the input data set further includes weather data measured at the one or more times prior to the prediction time.

### Claim 18 (depends on 15)

The method of claim 15 further comprising:
training the regression function on the history of demand for the resource, wherein the training is performed by the electronic data processing device and utilizes the baseline model of the demand.

### Claim 19 (depends on 18)

The method of claim 18 wherein the regression function comprises a support vector regression (SVR) function.

### Claim 20 (depends on 15)

The method of claim 15 wherein the baseline model comprises a Fourier model and modeling comprises:
computing Fourier components of the history of demand for the resource.