Predictive scoring based on key performance indicators in telecommunications system
A method includes: receiving protocol event data from a plurality of probes within the telecommunication system; determining a most probable cause of a call event from the protocol event data; applying the most probable cause to a trained machine learning algorithm that includes the most probable cause as its input and a telecommunication system score as its output; and in response to an output score from the trained machine learning algorithm, performing a corrective action for a plurality of network users that are expected to be affected by the most probable cause.
1. A method performed by a computing system that collects information in a telecommunication system, the method comprising:
receiving protocol event data from a plurality of probes within the telecommunication system;
determining a most probable cause of a network failure from the protocol event data;
training a machine learning algorithm by inputting historic net promoter score surveys and most probable cause data, thereby generating a trained machine learning algorithm that relates most probable causes of network failures to expected net promoter scores;
applying the most probable cause to the trained machine learning algorithm, which receives the most probable cause as input and provides a net promoter score as output based on the most probable cause; and
in response to the output net promoter score from the trained machine learning algorithm, performing a corrective action for a plurality of network users that are expected to correspond to the output net promoter score, the corrective action including using the most probable cause as a key to search through call records to identify the plurality of network users, and sending a message to the plurality of network users to apprise them of the most probable cause.
2. The method of claim 1 , further comprising sending a survey to a subset of the plurality of network users and verifying the output net promoter score using returned survey results.
3. The method of claim 1 , further comprising sending a survey to a subset of the plurality of network users and tuning the trained machine learning algorithm using returned survey results.
4. The method of claim 1 , wherein the trained machine learning algorithm includes an item selected from the list consisting of:
Naïve Bayes;
Deep Learning;
Random Forest; and
Gradient Boost.
5. The method of claim 1 , wherein the probes collect data being transmitted between a Radio Access Network (RAN) and a Mobility Management Element (MME).
6. The method of claim 1 , wherein the probes collect data being transmitted between a Radio Access Network (RAN) and a Serving Gateway (SGW).
7. The method of claim 1 , wherein the network failure is associated with a call event that includes a Key Performance Indicator (KPI) associated with call failure, an anomaly, or call degradation.
8. The method of claim 1 , wherein performing the corrective action for the plurality of network users also comprises an item selected from the list consisting of:
providing the plurality of network users with discounted use of the telecommunication system; and
providing the plurality of network users with additional data use within the telecommunication system.
9. A system comprising:
a processor; and
a memory having machine readable instructions that when executed by the processor, cause the system to:
receive protocol event data from a plurality of probes within a telecommunication system;
determine a most probable cause of a network failure from the protocol event data;
train a machine learning algorithm by inputting historic net promoter score surveys and most probable cause data, thereby generating a trained machine learning algorithm that relates most probable causes of network failures to expected net promoter scores;
apply the most probable cause to the trained machine learning algorithm, which receives the most probable cause as input and provides a net promoter score as output based on the most probable cause; and
in response to the output net promoter score from the trained machine learning algorithm, perform a corrective action for a plurality of network users that are expected to correspond to the output net promoter score, the corrective action including using the most probable cause as a key to search through call records to identify the plurality of network users, and sending a message to the plurality of network users to apprise them of the most probable cause.
10. The system of claim 9 , further comprising instructions to cause the system to send a survey to a subset of the plurality of network users and verifying the output net promoter score using returned survey results.
11. The system of claim 9 , further comprising instructions to cause the system to send a survey to a subset of the plurality of network users and tuning the trained machine learning algorithm using returned survey results.
12. The system of claim 9 , wherein the trained machine learning algorithm includes an item selected from the list consisting of:
Naïve Bayes;
Deep Learning;
Random Forest; and
Gradient Boost.
13. The system of claim 9 , wherein the probes collect data transmitted between a Radio Access Network (RAN) and a Mobility Management Element (MME).
14. The system of claim 9 , wherein the probes collect data transmitted between a Radio Access Network (RAN) and a Serving Gateway (SGW).
15. The system of claim 9 , wherein the network failure is associated with a call event that includes a Key Performance Indicator (KPI) associated with call failure, an anomaly, or call degradation.
16. The system of claim 9 , wherein the message is a short message service (SMS) message.
17. The system of claim 16 , wherein the SMS message also indicates an action to take.
18. The method of claim 1 , wherein the message is a short message service (SMS) message.
19. The method of claim 18 , wherein the SMS message also indicates an action to take.