Systems and methods for selecting a machine learning model that predicts a subscriber network experience in a geographic area

Patent metadata

Field	Value
Publication number	US-12501288-B2
Application number	US-202217931615-A
Country	US
Kind code	B2
Filing date	Sep 13, 2022
Priority date	Sep 13, 2022
Publication date	Dec 16, 2025
Grant date	Dec 16, 2025

Abstract

A device may receive network data identifying key performance indicators associated with a base station and user equipment (UEs), UE data identifying UE data records and UE locations, and geographic data identifying a geographic area and features of the geographic area. The device may correlate the network data, the UE data, and the geographic data to generate correlated data. The device may process, the correlated data, with a plurality of machine learning models, to generate a corresponding plurality of results, and may evaluate the plurality of results, with prediction models, to generate a set of results. The device may compare classification cost function weighted predictions and the set of results to generate comparisons, and may select a machine learning model, for the geographic area and from the plurality of machine learning models, based on the comparisons. The device may implement the machine learning model for the geographic area.

First claim

What is claimed is:

1 . A method, comprising:
receiving, by a device, network data identifying key performance indicators associated with a base station and user equipment provided in a geographic area; receiving, by the device, user equipment data identifying user equipment data records and user equipment locations in the geographic area; receiving, by the device, geographic data identifying the geographic area and features of the geographic area; performing, by the device, data engineering on the network data, the user equipment data, and the geographic data to generate engineered network data, engineered user equipment data, and engineered geographic data, respectively; correlating, by the device, the engineered network data, the engineered user equipment data, and the engineered geographic data to generate correlated data; processing, by the device, the correlated data, with a plurality of machine learning models, to generate a corresponding plurality of results; evaluating, by the device, the plurality of results, with prediction models, to generate a set of results; comparing, by the device, classification cost function weighted predictions and the set of results to generate comparisons; selecting, by the device, a machine learning model, for the geographic area and from the plurality of machine learning models, based on the comparisons; and implementing, by the device, the machine learning model for the geographic area.

2 . The method of claim 1 , further comprising:
receiving current network data identifying current key performance indicators associated with the base station and the user equipment provided in the geographic area; receiving current user equipment data identifying current user equipment data records and current user equipment locations in the geographic area; processing the current network data and the current user equipment data, with the machine learning model, to generate one or more predictions for the geographic area; and perform one or more actions based on the one or more predictions.

3 . The method of claim 2 , wherein performing the one or more actions comprises one or more of:
causing a technician to be dispatched to service the base station; causing an autonomous vehicle to be dispatched to service the base station; causing a repair to be scheduled for the base station; modifying one or more parameters for the base station; providing a suggestion to modify one or more parameters for the base station; causing a part to be ordered for the base station; or retraining the machine learning model based on the one or more predictions.

4 . The method of claim 2 , wherein performing the one or more actions comprises:
identifying a problem with the base station based on the one or more predictions; determining one or more parameters, associated with the base station, to modify in order to correct the problem with the base station; and instructing the base station to modify the one or more parameters in order to correct the problem with the base station.

5 . The method of claim 1 , wherein performing the data engineering on the network data, the user equipment data, and the geographic data to generate the engineered network data, the engineered user equipment data, and the engineered geographic data comprises:
performing a data cleansing technique on the network data, the user equipment data, and the geographic data to generate the engineered network data, the engineered user equipment data, and the engineered geographic data.

6 . The method of claim 1 , wherein performing the data engineering on the network data, the user equipment data, and the geographic data to generate the engineered network data, the engineered user equipment data, and the engineered geographic data comprises:
performing a feature engineering technique on the network data, the user equipment data, and the geographic data to generate the engineered network data, the engineered user equipment data, and the engineered geographic data.

7 . The method of claim 1 , wherein performing the data engineering on the network data, the user equipment data, and the geographic data to generate the engineered network data, the engineered user equipment data, and the engineered geographic data comprises:
performing a binning technique on the network data, the user equipment data, and the geographic data to generate the engineered network data, the engineered user equipment data, and the engineered geographic data.

8 . A device, comprising:
one or more processors configured to:
receive network data identifying key performance indicators associated with a base station and user equipment provided in a geographic area,
wherein the key performance indicators include one or more of:
throughput associated with the base station and the user equipment during a time period,
latency associated with the base station and the user equipment during the time period, or
packet loss associated with the base station and the user equipment during the time period;

receive user equipment data identifying user equipment data records and user equipment locations in the geographic area;
receive geographic data identifying the geographic area and features of the geographic area;
perform data engineering on the network data, the user equipment data, and the geographic data to generate engineered network data, engineered user equipment data, and engineered geographic data, respectively;
correlate the engineered network data, the engineered user equipment data, and the engineered geographic data to generate correlated data;
train a plurality of machine learning models with the correlated data to generate a corresponding plurality of results;
evaluate the plurality of results, with prediction models, to generate a set of results;
compare classification cost function weighted predictions and the set of results to generate comparisons;
select a machine learning model, for the geographic area and from the plurality of machine learning models, based on the comparisons; and
implement the machine learning model for the geographic area.

9 . The device of claim 8 , wherein the one or more processors, to correlate the engineered network data, the engineered user equipment data, and the engineered geographic data to generate the correlated data, are configured to:
process the engineered network data, the engineered user equipment data, and the engineered geographic data, with a correlation model, to generate the correlated data.

10 . The device of claim 8 , wherein the one or more processors, to evaluate the plurality of results, with the prediction models, to generate the set of results, are configured to:
evaluate the plurality of results, with a mean absolute error prediction model, to generate a first portion of the set of results; evaluate the plurality of results, with a standard deviation prediction model, to generate a second portion of the set of results; and combine the first portion and the second portion to generate the set of results.

11 . The device of claim 8 , wherein the one or more processors, to evaluate the plurality of results, with the prediction models, to generate the set of results, are configured to:
process the plurality of results, with the prediction models, to generate a ranked list of the plurality of results; and select the set of results based on the ranked list of the plurality of results.

12 . The device of claim 8 , wherein the comparisons include a ranked list of the set of results, and the one or more processors, to select the machine learning model, are configured to:
select the machine learning mod

Assignees

• Verizon Patent & Licensing Inc

Inventors

• Coyle Timothy E
• Liske Deborah Lynn
• Kapala Matthew
• Rehman Syed

Classifications

• H04W24/10 Primary

  Scheduling measurement reports {; Arrangements for measurement reports} · CPC title

• H04W24/02

  Arrangements for optimising operational condition · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06N20/10

  using kernel methods, e.g. support vector machines [SVM] · CPC title