# US 10,769,540 B2 · Rare event prediction

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

- Number: US 10,769,540 B2
- Document ID: US10769540B2
- Status: Granted
- Kind: B2
- Application number: 15/499,006
- Filed: 2017-04-27
- Published: 2018-11-01
- Granted: 2020-09-08
- Earliest claimed: 2017-04-27
- Assignees: Hewlett Packard Enterprise Development LP
- Applicants: Hewlett Packard Enterprise Development LP
- Inventors: Ajeet Subramanian; Debashree Ghosh; Swati Gupta
- Web page: https://patentgrep.com/patent/US10769540B2

## CPC

- G06N 5/04: Inference or reasoning models
- G06F 16/313: {Selection or weighting of terms for indexing}
- G06F 40/279: Recognition of textual entities
- G06N 20/00: Machine learning
- 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)
- G06Q 30/016: After-sales
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- 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

## Abstract

An example method for building a model to predict rare events is disclosed. The example disclosed herein comprises receiving a plurality of historical input logs wherein each log includes at least one key variable and unstructured data. The example further comprises applying text mining techniques to the unstructured data to obtain at least one predictor based on the unstructured data. The example further comprises creating a rare events prediction model based on the at least one key variable and the at least one predictor.

## Claims

### Claim 1

A method comprising:
receiving, by a processor, a plurality of historical input logs, wherein each log includes at least one key variable and unstructured data;
applying, by the processor, text mining techniques to the unstructured data to obtain at least one predictor based on the unstructured data;
selecting, from the plurality of historical input logs, a first sample of historical key variables and corresponding predictors;
oversampling the selected first sample into a plurality of simulated samples;
calculating escalation prediction model regression coefficients based on the plurality of simulated samples; and
creating, by the processor, an escalation prediction model based on the escalation prediction model regression coefficients, the at least one key variable, and the at least one predictor.

### Claim 2 (depends on 1)

The method of claim 1, the method further comprising:
receiving, by the processor, a new input log;
identifying, by the processor, at least one key variable and unstructured data from the new input log;
applying, by the processor, text mining techniques to the unstructured data from the new input log to obtain at least one predictor from the new input log; and
forecasting an escalation based on the escalation prediction model, the at least one key variable from the new input log, and the at least one predictor obtained from the new input log.

### Claim 3 (depends on 1)

The method of claim 1, wherein the unstructured data includes at least one of:
product features, industry of the customer, an explanation of the problem, customer comments, or a combination thereof.

### Claim 4 (depends on 1)

The method of claim 1, wherein applying text mining techniques comprises:
parsing the unstructured data into a plurality of parsed words;
removing stop words from the plurality of parsed words, wherein stop words are frequently used words that do not provide predictive value; and
stemming the plurality of parsed words into a plurality of predictors.

### Claim 5 (depends on 4)

The method of claim 4, further comprising identifying the stop words using string techniques.

### Claim 6 (depends on 4)

The method of claim 4, wherein stemming comprises grouping together parsed words that share the same root as a predictor.

### Claim 7 (depends on 1)

The method of claim 1, further comprising unbiasing the escalation prediction model regression coefficients to compensate for the oversampling.

### Claim 8 (depends on 1)

The method of claim 1, further validating the escalation prediction model based on a second sample of historical key variables and corresponding predictors different to the first sample of historical key variables and corresponding predictors.

### Claim 9

A system for predicting escalations in a customer service center the system comprising:
a physical processor;
a non-transitory storage medium storing machine readable instructions to cause the processor to:
receive a plurality of historical customer case logs, each log including a plurality of fields;
obtain a plurality of predictors from data included in the plurality of fields;
select a first sample of the customer case logs;
oversample the selected first sample into a plurality of simulated samples;
calculate escalation prediction model regression coefficients for the plurality of predictors based on the plurality of simulated samples; and
build an escalation prediction model based on the escalation prediction model regression coefficients.

### Claim 10 (depends on 9)

The system of claim 9, wherein the machine readable instructions further include instructions to cause the processor to:
receive a new customer case log;
obtain a second plurality of predictors from the new customer case log; and
forecast an escalation based on the escalation prediction model and the second plurality of predictors obtained from the new customer case log.

### Claim 11 (depends on 9)

The system of claim 9, wherein the plurality of fields include unstructured text data and the machine readable instructions include instructions to text mine the unstructured text data to obtain the plurality of predictors.

### Claim 12 (depends on 11)

The system of claim 11, wherein the machine readable instructions further include instructions to cause the processor to:
parse the unstructured text data into a plurality of parsed words;
remove stop words from the plurality of parsed words, wherein stop words are frequently used words that do not provide predictive value; and
stem the plurality of parsed words into a plurality of predictors.

### Claim 13 (depends on 12)

The system of claim 12, wherein the machine readable instructions further include instructions to cause the processor to identify the stop words using string techniques.

### Claim 14 (depends on 12)

The system of claim 12, wherein the machine readable instructions further comprise instructions to cause the processor to stem by grouping together parsed words that share a same root as a predictor.

### Claim 15 (depends on 9)

The system of claim 9, wherein the machine readable instructions further include instructions to cause the processor to unbias the escalation prediction model regression coefficients to compensate for the oversampling.

### Claim 16

A non-transitory machine-readable medium storing machine-readable instructions executable by a physical processor that cause the processor to:
receive a plurality of historical customer case logs, wherein each log includes at least one key variable and unstructured data;
parse the unstructured data into a plurality of parsed words;
remove stop words from the plurality of parsed words, wherein stop words are frequently used words that do not provide predictive value;
stem the plurality of parsed words into a plurality of predictors;
create an escalation prediction model based on the at least one key variable and the plurality of predictors;
receive a new customer case log;
identify at least one key variable and unstructured data from the new customer case log;
apply text mining techniques to the unstructured data from the new customer case log to obtain a corresponding plurality of predictors from the new customer case log; and
forecast an escalation to result from the new customer case log based on the escalation prediction model, the key variables from the new customer case log, and the plurality of predictors from the new customer case log.

### Claim 17 (depends on 16)

The non-transitory machine-readable medium of claim 16 further comprising machine readable instructions that are executable by the processor to:
select a first sample of historical key variables and corresponding predictors;
oversample the selected first sample into a plurality of simulated samples;
calculate escalation prediction model regression coefficients based on the plurality of simulated samples; and
build the escalation prediction model based on the escalation prediction model regression coefficients.

### Claim 18 (depends on 17)

The non-transitory machine-readable medium of claim 17 further comprising machine readable instructions that are executable by the processor to unbias the escalation prediction model regression coefficients to compensate for the oversampling.