System and method for identifying business logic and data lineage with machine learning

Patent metadata

Field	Value
Publication number	US-2021406222-A1
Application number	US-202117467938-A
Country	US
Kind code	A1
Filing date	Sep 7, 2021
Priority date	Aug 30, 2017
Publication date	Dec 30, 2021
Grant date	—

Abstract

An embodiment of the present invention is directed to implementing machine learning to define business logic and lineage. The system analyzes data patterns of SORs as well as consumption attributes to define the business logic. An embodiment of the present invention may achieve over 95% match rate for complex attributes. When provided with thousands of SOR attributes, the innovative system may identify a handful of relevant SOR attributes required as well as the business logic to derive the consumption attribute.

First claim

What is claimed is:

1 . A system that generates pseudo code that represents data logic from a source system to a target system, the system comprising:
a computer server comprising a programmed computer processor configured to perform the steps of: preprocessing source data using direct SQL and creating a comma separated values (CSV) file with header columns and target columns; processing the CSV file using dataframes; identifying a set of best source feature attributes using recursive feature elimination method in machine learning; separating the attributes to continuous and categorical columns; feeding the attributes to a machine learning algorithm; and generating a descriptive tree path in pseudo code.

2 . The system of claim 1 , wherein the pseudo code further comprises a series of IF THEN statements.

3 . The system of claim 1 , wherein the pseudo code is displayed on an interface user interface and automatically executed in the target system.

4 . The system of claim 1 , wherein the machine learning algorithm comprises a decision tree algorithm, a regression algorithm, or a Gaussian algorithm.

5 . The system of claim 1 , wherein the CSV file is processed using Pandas dataframes.

6 . The system of claim 1 , wherein the recursive feature elimination method includes iteratively constructing a model and choosing the best performing feature from each model construction based on coefficients.

7 . A method that generates pseudo code that represents data logic from a source system to a target system, the method comprising the steps of:
preprocessing, via a programmed computer processor, source data using direct SQL and creating a create comma separated values (CSV) file with header columns and target columns; processing, via the programmed computer processor, the CSV file using dataframes; identifying, via the programmed computer processor, a set of best source feature attributes using recursive feature elimination method in machine learning; separating, via the programmed computer processor, the attributes to continuous and categorical columns; feeding, via the programmed computer processor, the attributes to a machine learning algorithm; and generating, via the programmed computer processor, a descriptive tree path in pseudo code.

8 . The system of claim 1 , wherein the pseudo code further comprises a series of IF THEN statements.

9 . The system of claim 1 , wherein the pseudo code is displayed on an interface user interface and automatically executed in the target system.

10 . The system of claim 1 , wherein the machine learning algorithm comprises a decision tree algorithm, a regression algorithm, or a Gaussian algorithm.

11 . The system of claim 1 , wherein the CSV file is processed using Pandas dataframes.

12 . The system of claim 1 , wherein the recursive feature elimination method includes iteratively constructing a model and choosing the best performing feature from each model construction based on coefficients.

13 . A non-transient computer readable medium containing program instructions for causing a computer to perform a method that generates pseudo code that represents data logic from a source system to a target system, the method comprising the steps of:
preprocessing, via a programmed computer processor, source data using direct SQL and creating a create comma separated values (CSV) file with header columns and target columns; processing, via the programmed computer processor, the CSV file using dataframes; identifying, via the programmed computer processor, a set of best source feature attributes using recursive feature elimination method in machine learning; separating, via the programmed computer processor, the attributes to continuous and categorical columns; feeding, via the programmed computer processor, the attributes to a machine learning algorithm; and generating, via the programmed computer processor, a descriptive tree path in pseudo code.

14 . The system of claim 1 , wherein the pseudo code further comprises a series of IF THEN statements.

15 . The system of claim 1 , wherein the pseudo code is displayed on an interface user interface and automatically executed in the target system.

16 . The system of claim 1 , wherein the machine learning algorithm comprises a decision tree algorithm, a regression algorithm, or a Gaussian algorithm.

17 . The system of claim 1 , wherein the CSV file is processed using Pandas dataframes.

18 . The system of claim 1 , wherein the recursive feature elimination method includes iteratively constructing a model and choosing the best performing feature from each model construction based on coefficients.

Assignees

• Jpmorgan Chase Bank Na

Inventors

• Singh Jagmohan
• Vakalapudi Suneetha
• Tannehill Soren

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N5/046

  Forward inferencing; Production systems · CPC title

• G06F16/252 Primary

  between a Database Management System and a front-end application · CPC title