Detecting communication network insights of alerts

1. A method comprising, by a computing system:

identifying one or more geographic areas covered by a communication network;

determining, for each identified geographic area, a network performance metric for the identified geographic area based at least on a difference between:

(1) a first average network speed of the communication network in the identified geographic area during one or more first prior time periods in which the communication network is busy, and

(2) a second average network speed of the communication network in the identified geographic area during one or more second prior time periods in which the communication network is not busy;

comparing the respective performance metrics of the one or more geographic areas to a threshold network performance metric, wherein the threshold network performance metric is determined by a congestion-analysis machine learning (ML) model; and

identifying one or more network traffic congestions in one or more of the identified geographic areas having a determined network performance metric below the threshold network performance metric.

2. The method of claim 1 , wherein the network performance metric is based on a ratio of the different between the first and second average network speeds to the second average network speed.

3. The method of claim 1 , wherein the communication network has a network traffic volume being above a busy-hour threshold traffic volume during the one or more first prior time periods, and wherein the communication network has a second network traffic volume being below a non-busy-hour threshold traffic volume.

4. The method of claim 1 , further comprising:

sending one or more alerts to an operator upon identifying the one or more network traffic congestions in the one or more areas.

5. The method of claim 4 , further comprising:

ranking the one or more areas based on a congestion severity indicated by the respective performance metrics of the one or more areas; and

sending the one or more alerts with the one or more areas ranked based on the performance metrics.

6. The method of claim 1 , further comprising:

optimizing a network performance to increase a capacity of the communication network in the one or more geographic areas, wherein the network performance is optimized based on one or more performance parameters by aggregating a plurality of carriers.

7. The method of claim 1 , wherein the performance parameters comprise one or more of:

a load balancing factor;

a handover; or

a throughput.

8. The method of claim 1 , further comprising:

determining a correlation between a network performance degradation and a network traffic status or a network traffic status change.

9. The method of claim 1 , further comprising:

determining a correlation between a quality of experience metric and a network demand metric; and

predicting a time for a future network traffic congestion in the one or more geographic areas based on the correlation of the quality of experience metric and the network demand metric.

10. The method of claim 9 , wherein the quality of experience metric comprises a network download speed, and wherein the network demand metric comprises a request number of download requests.

11. The method of claim 1 , further comprising:

comparing a network traffic volume in the one or more geographic areas to an average network traffic volume in the one or more geographic areas; and

determining a temporal domain anomalous event when the network traffic volume is higher than the average network traffic volume by a threshold multiplying factor.

12. The method of claim 11 , wherein the first and second average network speeds are based on network traffic data points per hour per each individual day of a plurality of days.

13. The method of claim 1 , wherein the first and second average network speeds are based on network traffic data points per hour aggregating all days of a plurality of days.

14. The method of claim 13 , wherein the first average network speed is a first mean network speed in the one or more first prior time periods based on a plurality of aggregated network speeds in the one or more first prior time periods, and wherein each aggregated speed in the one or more first prior time periods is a second mean network speed in the plurality of days.

15. The method of claim 13 , wherein the first average network speed is a first median network speed in the one or more first prior time periods based on a plurality of aggregated network speeds in the one or more first prior time periods, and wherein each aggregated network speed in the one or more first prior time periods is a second median network speed in the plurality of days.

16. The method of claim 13 , wherein the first average network speed is a mean network speed in the one or more first time periods, and wherein the second average network speed is a mean network speed in one or more second time periods.

17. The method of claim 13 , wherein the first average network speed is a median network speed in the one or more first prior time periods, and wherein the second average network speed is a median network speed in one or more second prior time periods.

18. The method of claim 1 , wherein the threshold network performance metric is determined by a classification model or a tree model associated with the congestion-analysis machine leaning model trained by manually labeled data, and wherein the threshold network performance metric is adjusted by balancing a precision metric and a recall rate of the identified one or more network traffic congestions.

19. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:

identify one or more geographic areas covered by a communication network;

determine, for each identified geographic area, a network performance metric for the identified geographic area based at least on a difference between:

(1) a first average network speed of the communication network in the identified geographic area during one or more first prior time periods in which the communication network is busy, and

(2) a second average network speed of the communication network in the identified geographic area during one or more second prior time periods in which the communication network is not busy;

compare the respective performance metrics of the one or more geographic areas to a threshold network performance metric, wherein the threshold network performance metric is determined by a congestion-analysis machine learning (ML) model; and

identify one or more network traffic congestions in one or more of the identified geographic areas having a determined network performance metric below the threshold network performance metric.

20. A system comprising:

one or more non-transitory computer-readable storage media embodying instructions; and

one or more processors coupled to the storage media and operable to execute the instructions to:

identify one or more geographic areas covered by a communication network;

determine, for each identified geographic area, a network performance metric for the identified geographic area based at least on a difference between:

(1) a first average network speed of the communication network in the identified geographic area during one or more first prior time periods in which the communication network is busy, and

(2) a second average network speed of the communication network in the identified geographic area during one or more second prior time periods in which the communication network is not busy;

compare the respective performance metrics of the one or more geographic areas to a threshold network performance metric, wherein the threshold network performance metric is determined by a congestion-analysis machine learning (ML) model; and

identify one or more network traffic congestions in one or more of the identified geographic areas having a determined network performance metric below the threshold network performance metric.