Anomaly detection for bill generation
A bill verification system for verifying bill records associated with an entity account is disclosed. The system receives a request to verify a bill record associated with an entity account. The system identifies a bill level model to be used for verifying the bill record and detects based on the bill level model, bill level anomaly information for the bill record. The system additionally identifies a bill line level model to be used for verifying one or more bill lines in the bill record and detects based on the bill line level model, bill line level anomaly information for the bill record. The system then aggregates the bill level anomaly information and the bill line level anomaly information to generate a bill verification report for the bill record. The system provides the bill verification report as a response to the request received to verify the bill record.
1. A method comprising:
receiving, by a bill processing system implemented using one or more computing systems, a request to verify a bill record associated with an entity account, the bill record stored in a memory of the bill processing system;
identifying, by the bill processing system, a bill level machine learning (ML) model to be used for verifying the bill record, the bill level ML model trained using a training algorithm and historical data comprising historical bills and associated bill level features;
detecting, by the bill processing system, based at least in part on the bill level ML model, bill level anomaly information for the bill record based on a spatial distribution of the historical data learned by the bill level model using the training algorithm and the bill record, wherein the bill level anomaly information comprises information about an anomaly detected in the bill record;
identifying, by the bill processing system, a bill line level ML model to be used for verifying one or more bill lines in the bill record, the bill line level ML model trained using the training algorithm, the historical data comprising the historical bills and associated bill line level features, wherein the training algorithm includes a supervised learning training algorithm or an unsupervised learning training algorithm;
detecting, by the bill processing system, based at least in part on the bill line level ML model, bill line level anomaly information for the bill record based on the spatial distribution of the historical data learned by the bill line level model using the training algorithm and the bill record, wherein the bill line level anomaly information comprises information about an anomaly detected at a bill line in the bill record;
receiving, by the bill processing system, a vector representation for verifying the bill record, the vector representation identifying a set of products that are expected to be included in a billing cycle of the bill record;
identifying, by the bill processing system, additional anomaly information associated with the bill record based on the vector representation;
aggregating, by the bill processing system, the bill level anomaly information detected by the bill level ML model, the bill line level anomaly information detected by the bill line level ML model, and the additional anomaly information to generate a bill verification report for the bill record;
providing, by the bill processing system, the bill verification report as a response to the request received to verify the bill record, the bill verification report identifying the additional anomaly information associated with the bill record, wherein the additional anomaly information identifies a product that is not in the set of products identified in the vector representation;
analyzing, by a feedback system that is communicatively coupled to the bill processing system, the bill verification report for the bill record; and
providing, by the feedback system, a result of the analysis, in real-time, to a training system used by the bill processing system to train a set of ML models for verifying a set of bill records associated with a set of entity accounts, wherein the set of ML models include at least the bill level ML model to be used for verifying the set of bill records and the bill line level ML model to be used for verifying the set of bill records.
2. The method of claim 1 , wherein:
the additional anomaly information identifies at least one of one or more missing bill lines in the bill record or one or more incorrectly added bill lines in the bill record.
3. The method of claim 2 , wherein the vector representation is a count of a set of products subscribed to by a set of users associated with the entity account.
4. The method of claim 1 , wherein the bill level anomaly information further comprises information about a cause of the detected anomaly and if the detected anomaly was caused due to a seasonal variation.
5. The method of claim 1 , wherein the bill line level anomaly information further comprises information about a cause of the detected anomaly at the bill line and if the detected anomaly at the bill line was caused due to a seasonal variation, wherein the bill line identifies a product or a service used by the entity account.
6. The method of claim 1 , wherein identifying, by the bill processing system, the bill level ML model to be used for verifying the bill record comprises:
identifying a set of entity accounts that are similar to the entity account associated with the bill record; and
identifying the bill level ML model to be used for verifying the bill record based on the set of entity accounts.
7. The method of claim 1 , wherein identifying, by the bill processing system, the bill line level ML model to be used for verifying the bill record comprises:
identifying a set of entity accounts that are similar to the entity account associated with the bill record; and
identifying the bill line level ML model to be used for verifying the bill record based on the set of entity accounts.
8. The method of claim 7 , wherein the set of entity accounts are identified by clustering one or more entity accounts based on usage patterns, product subscriptions, regions, usage volumes and billing amount ranges in a set of bill records processed by the bill processing system for the one or more entity accounts.
9. The method of claim 1 , wherein the entity account represents at least one of an organization that subscribes to a set of services provided by the bill processing system for performing bill verification or an individual user of the organization.
10. A bill processing system comprising:
a memory; and
one or more processors configured to perform processing, the processing comprising:
receiving a request to verify a bill record associated with an entity account, the bill record stored in the memory of the bill processing system;
identifying a bill level machine learning (ML) model to be used for verifying the bill record, the bill level ML model trained using a training algorithm and historical data comprising historical bills and associated bill level features;
detecting based at least in part on the bill level ML model, bill level anomaly information for the bill record based on a spatial distribution of the historical data learned by the bill level model using the training algorithm and the bill record, wherein the bill level anomaly information comprises information about an anomaly detected in the bill record;
identifying a bill line level ML model to be used for verifying one or more bill lines in the bill record, the bill line level ML model trained using the training algorithm, the historical data comprising the historical bills and associated bill line level features, wherein the training algorithm includes a supervised learning training algorithm or an unsupervised learning training algorithm;
detecting based at least in part on the bill line level ML model, bill line level anomaly information for the bill record based on a spatial distribution of the historical data learned by the bill line level model using the training algorithm and the bill record, wherein the bill line level anomaly information comprises information about an anomaly detected at a bill line in the bill record;
receiving, by the bill processing system, a vector representation for verifying the bill record, the vector representation identifying a set of products that are expected to be included in a billing cycle of the bill record;
identifying, by the bill processing system, additional anomaly information associated with the bill record based on the vector representation;
aggregating the bill level anomaly information detected by the bill level ML model, the bill line level anomaly information detected by the bill line level ML model, and the additional anomaly information to generate a bill verification report for the bill record;
providing the bill verification report as a response to the request received to verify the bill record, the bill verification report identifying the additional anomaly information associated with the bill record, wherein the additional anomaly information identifies a product that is not in the set of products identified in the vector representation;
analyzing, by a feedback system that is communicatively coupled to the bill processing system, the bill verification report for the bill record; and
providing, by the feedback system, a result of the analysis, in real-time, to a training system used by the bill processing system to train a set of ML models for verifying a set of bill records associated with a set of entity accounts, wherein the set of ML models include at least the bill level ML model to be used for verifying the set of bill records and the bill line level ML model to be used for verifying the set of bill records.
11. The system of claim 10 , wherein the additional anomaly information identifies at least one of one or more missing bill lines in the bill record or one or more incorrectly added bill lines in the bill record.
12. The system of claim 11 , wherein the vector representation is a count of a set of products subscribed to by a set of users associated with the entity account.
13. The system of claim 10 , wherein the bill level anomaly information further comprises information about a cause of the detected anomaly and if the detected anomaly was caused due to a seasonal variation.
14. The system of claim 10 , wherein the bill line level anomaly information further comprises information about a cause of the detected anomaly at the bill line and if the detected anomaly at the bill line was caused due to a seasonal variation, wherein the bill line identifies a product or a service used by the entity account.
15. A non-transitory computer-readable medium having program code that is stored thereon, the program code executable by one or more processing devices for performing operations comprising:
receiving a request to verify a bill record associated with an entity account, the bill record stored in a memory of the one or more processing devices;
identifying a bill level machine learning (ML) model to be used for verifying the bill record, the bill level ML model trained using a training algorithm and historical data comprising historical bills and associated bill level features;
detecting based at least in part on the bill level ML model, bill level anomaly information for the bill record based on a spatial distribution of the historical data learned by the bill level model using the training algorithm and the bill record, wherein the bill level anomaly information comprises information about an anomaly detected in the bill record;
identifying a bill line level ML model to be used for verifying one or more bill lines in the bill record, the bill line level ML model trained using the training algorithm, the historical data comprising the historical bills and associated bill line level features, wherein the training algorithm includes a supervised learning training algorithm or an unsupervised learning training algorithm;
detecting based at least in part on the bill line level ML model, bill line level anomaly information for the bill record based on a spatial distribution of the historical data learned by the bill line level model using the training algorithm and the bill record, wherein the bill line level anomaly information comprises information about an anomaly detected at a bill line in the bill record;
receiving, by the bill processing system, a vector representation for verifying the bill record, the vector representation identifying a set of products that are expected to be included in a billing cycle of the bill record;
identifying, by the bill processing system, additional anomaly information associated with the bill record based on the vector representation;
aggregating the bill level anomaly information detected by the bill level ML model, the bill line level anomaly information detected by the bill line level ML model, and the additional anomaly information to generate a bill verification report for the bill record; and
providing the bill verification report as a response to the request received to verify the bill record, the bill verification report identifying the additional anomaly information associated with the bill record, wherein the additional anomaly information identifies a product that is not in the set of products identified in the vector representation;
analyzing, by a feedback system that is communicatively coupled to the bill processing system, the bill verification report for the bill record; and
providing, by the feedback system, a result of the analysis, in real-time, to a training system used by the bill processing system to train a set of ML models for verifying a set of bill records associated with a set of entity accounts, wherein the set of ML models include at least the bill level ML model to be used for verifying the set of bill records and the bill line level ML model to be used for verifying the set of bill records.
16. The non-transitory computer-readable medium of claim 15 , wherein identifying a bill level ML model to be used for verifying the bill record comprises:
identifying a set of entity accounts that are similar to the entity account associated with the bill record; and
identifying the bill level ML model to be used for verifying the bill record based on the set of entity accounts.
17. The non-transitory computer-readable medium of claim 15 , wherein the entity account represents at least one of an organization that subscribes to a set of services provided by a bill processing system for performing bill verification or an individual user of the organization.