Wage garnishments processing using machine learning for predicting field values

1. A computer-implemented method for processing wage garnishment court orders, the method comprising:

obtaining a court order for wage garnishment for a user from a plurality of users;

predicting values for a first subset of a number of fields for executing wage garnishment according to the court order using a rule engine;

training a machine learning engine comprises a plurality of machine learning models with a plurality of classification algorithms, wherein each machine learning model from the plurality of machine learning models comprises a different classification algorithm from the plurality of classification algorithms;

predicting values for a second subset of the number of fields using a machine learning model from the machine learning engine, wherein the second subset of the number of fields comprises missing fields for which the rule engine could not provide a prediction, and wherein the machine learning model selected to use for predicting field values is selected according to at least one of accuracy, speed, size of model, and explainability;

automatically receiving executed court orders for wage garnishment for the plurality of users in a period of time;

extracting lien features and client profile features from the executed court orders received for the plurality of users in the period of time; and

periodically retraining the machine learning engine using the lien features and client profile features from the executed court orders for wage garnishment received for the plurality of users in the period of time.

2. The method of claim 1 , wherein the plurality of classification algorithms comprises a decision tree algorithm, a random forest algorithm, a Naïve Bayes algorithm, and a K-nearest neighbor algorithm.

3. The method of claim 1 , further comprising:

performing a comparative analysis of the plurality of classification algorithms to select the one of the plurality of classification algorithms selected to use for the machine learning engine for predicting field values.

4. The method of claim 1 , wherein the machine learning engine is trained from historic data.

5. The method of claim 1 , further comprising:

saving an audit of changed field values after rule engine processing and machine learning engine processing.

6. The method of claim 5 , further comprising:

comparing a respective audit record of field value predictions from the rule engine and from the machine learning engine with final values to determine an accuracy of the predictions from the rule engine and from the machine learning engine.

7. The method of claim 1 , further comprising:

performing initial training of the machine learning engine from manual training; and

re-training the machine learning engine periodically without manual training.

8. The method of claim 1 , further comprising:

determining whether to use machine learning engine according to a flag which indicates whether the machine learning engine has been trained such that the accuracy of the predicted values from the machine learning engine exceed a threshold accuracy.

9. The method of claim 8 , wherein manual entry of field values for which the rule engine failed to provide a prediction is performed until the machine learning engine has been trained.

10. The method of claim 1 , wherein the rule engine comprises a Drools rule engine.

11. A system for processing wage garnishment court orders, the system comprising:

a bus system;

a storage device connected to the bus system, wherein the storage device stores program instructions; and

a number of processors connected to the bus system, wherein the number of processors execute the program instructions to:

obtain a court order for wage garnishment for a user from a plurality of users;

predict values for a first subset of a number of fields for executing wage garnishment according to the court order using a rule engine;

training a machine learning engine comprises a plurality of machine learning models with a plurality of classification algorithms, wherein each machine learning model from the plurality of machine learning models comprises a different classification algorithm from the plurality of classification algorithms;

predict values for a second subset of the number of fields using a machine learning model from the machine learning engine, wherein the second subset of the number of fields comprises missing fields for which the rule engine could not provide a prediction, and wherein the machine learning model selected to use for the machine learning engine for predicting field values is selected according to at least one of accuracy, speed, size of model, and explainability;

automatically receiving executed court orders for wage garnishment for the plurality of users in a period of time;

extracting lien features and client profile features from the executed court orders received for the plurality of users in the period of time; and

periodically retraining the machine learning engine using the lien features and client profile features from the executed court orders for wage garnishment received for the plurality of users in the period of time.

12. The system of claim 11 , wherein the plurality of classification algorithms comprises a decision tree algorithm, a random forest algorithm, a Naïve Bayes algorithm, and a K-nearest neighbor algorithm.

13. The system of claim 11 , wherein the number of processors further execute the program instructions to:

perform a comparative analysis of the plurality of classification algorithms to select the one of the plurality of classification algorithms selected to use for the machine learning engine for predicting field values.

14. The system of claim 11 , wherein the machine learning engine is trained from historic data.

15. The system of claim 11 , wherein the number of processors further execute the program instructions to:

save an audit of changed field values after rule engine processing and machine learning engine processing.

16. The system of claim 15 , wherein the number of processors further execute the program instructions to:

compare a respective audit record of field value predictions from the rule engine and from the machine learning engine with final values to determine an accuracy of the predictions from the rule engine and from the machine learning engine.

17. The system of claim 11 , wherein the number of processors further execute the program instructions to:

perform initial training of the machine learning engine from manual training; and

re-train the machine learning engine periodically without manual training.

18. The system of claim 11 , wherein the number of processors further execute the program instructions to:

determine whether to use machine learning engine according to a flag which indicates whether the machine learning engine has been trained such that the accuracy of the predicted values from the machine learning engine exceed a threshold accuracy.

19. The system of claim 18 , wherein manual entry of field values for which the rule engine failed to provide a prediction is performed until the machine learning engine has been trained.

20. The system of claim 11 , wherein the rule engine comprises a Drools rule engine.

21. A computer program product for processing wage garnishment court orders, the computer program product comprising:

a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a number of processors to cause a computer to perform the steps of:

obtaining a court order for wage garnishment for a user from a plurality of users;

predicting values for a first subset of a number of fields for executing wage garnishment according to the court order using a rule engine;

training a machine learning engine comprises a plurality of machine learning models with a plurality of classification algorithms, wherein each machine learning model from the plurality of machine learning models comprises a different classification algorithm from the plurality of classification algorithms;

predicting values for a second subset of the number of fields using a machine learning model from the machine learning engine, wherein the second subset of the number of fields comprises missing fields for which the rule engine could not provide a prediction, and wherein the machine learning model selected to use for predicting field values is selected according to at least one of accuracy, speed, size of model, and explainability;

automatically receiving executed court orders for wage garnishment for the plurality of users in a period of time;

extracting lien features and client profile features from the executed court orders received for the plurality of users in the period of time; and

periodically retraining the machine learning engine using the lien features and client profile features from the executed court orders for wage garnishment received for the plurality of users in the period of time.

22. The computer program product of claim 21 , wherein the plurality of classification algorithms comprises a decision tree algorithm, a random forest algorithm, a Naïve Bayes algorithm, and a K-nearest neighbor algorithm.

23. The computer program product of claim 21 , wherein the program instructions further comprise program instructions executable by a number of processors to cause the computer to perform the steps of:

performing a comparative analysis of the plurality of classification algorithms to select the one of the plurality of classification algorithms selected to use for the machine learning engine for predicting field values.

24. The computer program product of claim 21 , wherein the machine learning engine is trained from historic data.

25. The computer program product of claim 21 , wherein the program instructions further comprise program instructions executable by a number of processors to cause the computer to perform the steps of:

saving an audit of changed field values after rule engine processing and machine learning engine processing.

26. The computer program product of claim 25 , wherein the program instructions further comprise program instructions executable by a number of processors to cause the computer to perform the steps of:

comparing a respective audit record of field value predictions from the rule engine and from the machine learning engine with final values to determine an accuracy of the predictions from the rule engine and from the machine learning engine.

27. The computer program product of claim 21 , wherein the program instructions further comprise program instructions executable by a number of processors to cause the computer to perform the steps of:

performing initial training of the machine learning engine from manual training; and

re-training the machine learning engine periodically without manual training.

28. The computer program product of claim 21 , wherein the program instructions further comprise program instructions executable by a number of processors to cause the computer to perform the steps of:

determining whether to use machine learning engine according to a flag which indicates whether the machine learning engine has been trained such that the accuracy of the predicted values from the machine learning engine exceed a threshold accuracy.

29. The computer program product of claim 28 , wherein manual entry of field values for which the rule engine failed to provide a prediction is performed until the machine learning engine has been trained.

30. The computer program product of claim 21 , wherein the rule engine comprises a Drools rule engine.