Predicting whether a user of a wireless telecommunication network will report a problem

1. A method to predict whether a user of a wireless telecommunication network will report a problem associated with the wireless telecommunication network, the method comprising:

obtaining multiple key performance indicators (KPIs) describing a user experience with the wireless telecommunication network,

wherein the multiple KPIs comprise at least two of: a billing account number (BAN) count, how long the user has been using the wireless telecommunication network, Reference Signals Received Power (RSRP), and Reference Signal Received Quality (RSRQ);

calculating at least a daily value of each KPI among the multiple KPIs by processing a KPI among the multiple KPIs according to a rule specific to the KPI;

creating an image representing a color-coded value of each KPI among the multiple KPIs,

wherein a first axis of the image represents the KPI among the multiple KPIs, and

wherein a second axis of the image represents the daily value of the KPI among the multiple KPIs; and

predicting whether the user of the wireless telecommunication network will report the problem associated with the wireless telecommunication network by:

providing the image to a machine learning model, and

receiving a prediction from the machine learning model whether the user of the wireless telecommunication network will report the problem.

2. The method of claim 1 , further comprising:

obtaining multiple reports of multiple problems associated with the wireless telecommunication network, and multiple geolocations of the multiple reports;

identifying a geographic area among the multiple geolocations having a number of reports exceeding a predetermined threshold; and

flagging the geographic area as requiring remedial action.

3. The method of claim 1 , the machine learning model comprising a convolutional neural network.

4. The method of claim 1 , further comprising:

training the machine learning model by obtaining the multiple KPIs within a first time window;

obtaining the multiple KPIs within a second time window, wherein the first time window and the second time window do not overlap; and

using, by the machine learning model, the multiple KPIs within the second time window as an input to the machine learning model to predict whether the user of the wireless telecommunication network will report the problem.

5. The method of claim 1 , wherein processing the multiple KPIs further comprises:

comparing at a predetermined time period the KPI to a predetermined threshold;

computing a count of how many comparisons are below the predetermined threshold within a day; and

setting the daily value of the KPI to be equal to the count.

6. The method of claim 1 , wherein processing the multiple KPIs further comprises:

comparing at a predetermined time period the KPI to a predetermined threshold;

determining whether a predetermined number of consecutive comparisons produce the same result;

combining the consecutive comparisons into a single comparison;

computing a count of how many comparisons are below the predetermined threshold within a day; and

setting the daily value of the KPI to be equal to the count.

7. The method of claim 1 , wherein the multiple KPIs comprise at least three of an hourly throughput count, a software change count, a device change count, a customer complaint count, a ratio of dropped calls to placed calls, and a number of repeated calls.

8. At least one non-transient, computer-readable medium, carrying instructions that, when executed by at least one data processor, performs a method comprising:

obtaining multiple key performance indicators (KPIs) describing a user experience with a wireless telecommunication network;

calculating, at a first predetermined time period, a value of each KPI among the multiple KPIs by processing a KPI among the multiple KPIs according to a rule specific to the KPI;

creating an image representing a color-coded value of the KPI among the multiple KPIs,

wherein a first axis of the image represents the KPI among the multiple KPIs, and

wherein a second axis of the image represents the value of the KPI among the multiple KPIs; and

predicting whether a user of the wireless telecommunication network will report a problem associated with the wireless telecommunication network by:

providing the image to a machine learning model, and

receiving a prediction from the machine learning model whether the user of the wireless telecommunication network will report the problem.

9. The non-transient, computer-readable medium of claim 8 , further comprising:

obtaining multiple reports of multiple problems associated with the wireless telecommunication network, and multiple geolocations of the multiple reports;

identifying a geographic area among the multiple geolocations having a number of reports exceeding a predetermined threshold; and

marking the geographic area as requiring remedial action.

10. The non-transient, computer-readable medium of claim 8 , further comprising:

training the machine learning model by obtaining the multiple KPIs within a first time window;

obtaining the multiple KPIs within a second time window, wherein the first time window and the second time window do not overlap; and

using, by the machine learning model, the multiple KPIs within the second time window as an input to the machine learning model to predict whether the user of the wireless telecommunication network will report the problem.

11. The non-transient, computer-readable medium of claim 8 , wherein processing the multiple KPIs further comprises:

comparing at a second predetermined time period the KPI to a predetermined threshold;

computing a count of how many comparisons are below the predetermined threshold within the first predetermined time period; and

setting the value of the KPI to be equal to the count.

12. The non-transient, computer-readable medium of claim 8 , wherein processing the multiple KPIs further comprises:

comparing at a second predetermined time period the KPI to a predetermined threshold;

determining whether a predetermined number of consecutive comparisons produce the same result;

combining the consecutive comparisons into a single comparison;

computing a count of how many comparisons are below the predetermined threshold within the first predetermined time period; and

setting the value of the KPI to be equal to the count.

13. The non-transient, computer-readable medium of claim 8 , wherein the multiple KPIs comprise at least three of a billing account number (BAN) count, how long the user has been using the wireless telecommunication network, Reference Signals Received Power (RSRP), a Reference Signal Received Quality (RSRQ), an hourly throughput count, a software change count, a device change count, a customer complaint count, a ratio of dropped calls to placed calls, and a number of repeated calls.

14. A system comprising:

one or more processors; and

memory coupled to the one or more processors, wherein the memory includes instructions executable by the one or more processors to:

obtain multiple key performance indicators (KPIs) describing a user experience with a wireless telecommunication network;

calculate, at a first predetermined time period, a value of each KPI among the multiple KPIs by processing a KPI among the multiple KPIs according to a rule specific to the KPI;

create an image representing a color-coded value of the KPI among the multiple KPIs,

wherein a first axis of the image represents the KPI among the multiple KPIs, and

wherein a second axis of the image represents the value of the KPI among the multiple KPIs;

predict whether a user of the wireless telecommunication network will report a problem associated with the wireless telecommunication network by:

providing the image to a machine learning model, and

receiving a prediction from the machine learning model whether the user of the wireless telecommunication network will report the problem.

15. The system of claim 14 , further comprising the instructions to:

obtain multiple reports of multiple problems associated with the wireless telecommunication network, and multiple geolocations of the multiple reports;

identify a geographic area among the multiple geolocations having a number of reports exceeding a predetermined threshold; and

identify the geographic area as requiring remedial action.

16. The system of claim 14 , the machine learning model comprising a convolutional neural network.

17. The system of claim 14 , further comprising the instructions to:

train the machine learning model by obtaining the multiple KPIs within a first time window;

obtain the multiple KPIs within a second time window, wherein the first time window and the second time window do not overlap; and

use, by the machine learning model, the multiple KPIs within the second time window as an input to the machine learning model to predict whether the user of the wireless telecommunication network will report the problem.

18. The system of claim 14 , wherein the instructions to process the multiple KPIs further comprise the instructions to:

compare at a second predetermined time period the KPI to a predetermined threshold;

compute a count of how many comparisons are below the predetermined threshold within the first predetermined time period; and

set the value of the KPI to be equal to the count.

19. The system of claim 14 , wherein the instructions to process the multiple KPIs further comprise the instructions to:

compare at a second predetermined time period the KPI to a predetermined threshold;

determine whether a predetermined number of consecutive comparisons produce the same result;

combine the consecutive comparisons into a single comparison;

compute a count of how many comparisons are below the predetermined threshold within the first predetermined time period; and

set the value of the KPI to be equal to the count.

20. The system of claim 14 , wherein the multiple KPIs comprise at least three of a billing account number (BAN) count, how long the user has been using the wireless telecommunication network, Reference Signals Received Power (RSRP), a Reference Signal Received Quality (RSRQ), an hourly throughput count, a software change count, a device change count, a customer complaint count, a ratio of dropped calls to placed calls, and a number of repeated calls.