# US 2020/0401940 A1 · Framework for choosing the appropriate generalized linear model

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

- Number: US 2020/0401940 A1
- Document ID: US20200401940A1
- Status: Published application
- Kind: A1
- Application number: 16/897,400
- Filed: 2020-06-10
- Published: 2020-12-24
- Earliest claimed: 2019-06-18
- Assignees: Verint Americas Inc.
- Applicants: Verint Americas Inc.
- Inventors: Cynthia Freeman
- Web page: https://patentgrep.com/patent/US20200401940A1

## CPC

- 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)}
- G06F 18/2433: Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
- G06N 20/00: Machine learning
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Systems and methods are provided framework for automatically choosing the appropriate generalized linear model (GLM) given a time series of count data, and for anomaly detection on time series data. A dispersion parameter is determined and used to determine whether the count data is overdispersed data or underdispersed data. The overdispersed data or the underdispersed data is used to determine a GLM to apply on the dataset. Using the determined GLM on the data, anomalies can be determined.

## Claims

### Claim 1

A method for determining a generalized linear model (GLM) to apply on a dataset, the method comprising:
receiving a dataset comprising count data by a computing device;
determining whether the count data is overdispersed data or underdispersed data by the computing device; and
using the overdispersed data or the underdispersed data to determine a GLM to apply on the dataset by the computing device.

### Claim 2 (depends on 1)

The method of claim 1, wherein the count data is a time series of count data.

### Claim 3 (depends on 1)

The method of claim 1, further comprising determining when the variance of the count data exceeds the mean of the count data, the count data is overdispersed data, and when the mean of the count data exceeds the variance of the count data, the count data is underdispersed data.

### Claim 4 (depends on 1)

The method of claim 1, wherein determining whether the count data is overdispersed data or underdispersed data comprises using a dispersion parameter.

### Claim 5 (depends on 4)

The method of claim 4, wherein the dispersion parameter is equivalent to (χ2)/Df Residuals, wherein chi-squared (χ2) represents the sum of the squares of the residuals of the model, and Df Residuals represents the residual degrees of freedom n-p-1 where n is the number of observations and p is the number of regressors.

### Claim 6 (depends on 5)

The method of claim 5, further comprising determining the count data to be overdispersed data when the dispersion parameter is greater than 1, and determining the count data to be underdispersed data when the dispersion parameter is less than 1.

### Claim 7 (depends on 1)

The method of claim 1, wherein the GLM is determined from a group comprising a Poisson regression GLM, a generalized Poisson GLM, a negative binomial GLM, a zero-inflated Poisson GLM, and a hurdle Poisson GLM.

### Claim 8 (depends on 1)

The method of claim 1, further comprising applying the GLM on the dataset and outputting a result of the applying to a client device.

### Claim 9 (depends on 1)

The method of claim 1, wherein using the overdispersed data or the underdispersed data to determine the GLM comprises using a standard deviations threshold value, a confidence interval value, a zero count threshold value, a source count value, and a coefficient determining method indicator.

### Claim 10

A system for determining a generalized linear model (GLM) to apply on a dataset, the system comprising:
a dispersion determination module that determines whether the count data is overdispersed data or underdispersed data; and
a GLM selector module that uses the overdispersed data or the underdispersed data to determine a GLM to apply on the dataset.

### Claim 11 (depends on 10)

The system of claim 10, wherein the count data is a time series of count data.

### Claim 12 (depends on 10)

The system of claim 10, wherein when the variance of the count data exceeds the mean of the count data, the count data is overdispersed data, and when the mean of the count data exceeds the variance of the count data, the count data is underdispersed data.

### Claim 13 (depends on 10)

The system of claim 10, where the dispersion determination module determines a dispersion parameter, and wherein determining whether the count data is overdispersed data or underdispersed data comprises using the dispersion parameter.

### Claim 14 (depends on 13)

The system of claim 13, wherein the dispersion parameter is equivalent to (χ2)/Df Residuals, wherein chi-squared (χ2) represents the sum of the squares of the residuals of the model, and Df Residuals represents the residual degrees of freedom n-p-1 where n is the number of observations and p is the number of regressors.

### Claim 15 (depends on 14)

The system of claim 14, wherein the count data is determined to be overdispersed data when the dispersion parameter is greater than 1, and the count data is determined to be underdispersed data when the dispersion parameter is less than 1.

### Claim 16 (depends on 10)

The system of claim 10, wherein the GLM is determined from a group comprising a Poisson regression GLM, a generalized Poisson GLM, a negative binomial GLM, a zero-inflated Poisson GLM, and a hurdle Poisson GLM.

### Claim 17 (depends on 10)

The system of claim 10, further comprising a computing device that applies the GLM on the dataset and outputs a result of the applying to a client device.

### Claim 18 (depends on 10)

The system of claim 10, wherein using the overdispersed data or the underdispersed data to determine the GLM comprises using a standard deviations threshold value, a confidence interval value, a zero count threshold value, a source count value, and a coefficient determining method indicator.

### Claim 19

A method for determining an anomaly in a dataset, the method comprising:
receiving a dataset comprising a time series of count data by a computing device;
determining whether the count data is overdispersed data or underdispersed data by the computing device;
using the overdispersed data or the underdispersed data to determine a generalized linear model (GLM) to apply on the dataset by the computing device;
applying the GLM on the dataset to generate a result by the computing device; and
determining that an anomaly exists in the dataset using the result by the computing device.

### Claim 20 (depends on 19)

The method of claim 19, further comprising, prior to determining whether the count data is overdispersed data or underdispersed data:
applying a Poisson regression GLM on the dataset;
determining a dispersion parameter based on the results of the Poisson regression GLM;
determining the number of standard deviations away from 1 that the dispersion parameter is;
comparing the number of standard deviations to a threshold; and
determining the GLM to apply on the dataset using the results of the comparing and the dispersion parameter.