Systems and methods of assisted strategy design

• US 10,417,704 B2
• Filed: 01/05/2015
• Issued: 09/17/2019
• Est. Priority Date: 11/02/2010
• Status: Active Grant

• Current Assignee: Experian Technology Ltd. (Experian plc)
• Original Assignee: Experian Technology Ltd. (Experian plc)
• Inventors: Searson, Paul A., Hastings, Maria M.
• Primary Examiner(s): Tran, Hai
• Application Number:
  US14/589,250
• Publication Number:
  US 20150278944A1
• Time in Patent Office: 1,716 Days
• Field of Search: 705 38
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/045 Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03
• CPC Class Codes: G06Q 40/03 Credit; Loans; Processing thereof

Abstract

In an embodiment, a computer system performs a method of modifying a decision algorithm, such as an algorithm for assigning credit limits or other credit terms, via a graphical user interface. The system displays a graphical representation of a decision tree comprising a plurality of decision nodes. The decision nodes are associated with respective conditions, and at least a portion of the decision nodes are associated with an outcome to be assigned to an individual matching the associated respective conditions in accordance with the decision tree. The system receives a selection of a decision node and selects a parameter to be used to construct subnodes of the selected decision node, and constructs the subnodes. The system calculates an outcome for each subnode, and it updates the graphical representation of the decision tree to include the plurality of subnodes.

Claims

1. A computer system comprising:

one or more computer processors;

a computer-readable memory having stored thereon a plurality of executable instructions configured to be executed on the one or more computer processors to perform operations comprising;

obtaining information describing process steps included in a user-defined process flow, with one or more of the process steps being associated with a corresponding one or more decision trees usable to determine information associated with a consumer based on inputs describing parameters of the consumer;

providing, for presentation to a user, a graphical user interface depicting a decision tree comprising;

a plurality of nodes including a plurality of decision nodes and a plurality of leaf nodes;

wherein each decision node is associated with one or more leaf nodes, and each of the leaf nodes is associated with different value ranges of a parameter of consumers associated with the particular decision node;

receiving, from a user, user interaction requesting a next best split of a selected leaf node associated with a selected consumer parameter;

in response to the next best split request;

automatically, by the computing device, identifying a plurality of parameters of consumers usable to further segment consumers associated with the selected leaf node;

identifying, for each of the identified plurality of parameters, one or more leaf nodes associated with different value ranges of the respective parameter;

providing, for presentation to the user, an indication of the identified plurality of parameters as possible split decision nodes, including statistical parameters indicating respective rankings of the possible split decision nodes, wherein the rankings of the possible split decision nodes are determined automatically by the computing device;

receiving user input selecting a first of the possible split decision nodes to be merged into the decision tree;

incorporating, into the decision tree, the selected first possible decision node and the identified one or more leaf nodes associated with the selected first possible decision node

receiving user input associated with calculating proposed outcomes for association with the leaf nodes, the user input indicating constraints to be optimized; and

determining, based on the constraints, associations between proposed outcomes and leaf nodes,

wherein the updated decision tree is usable to segment consumers and determine associated outcomes for the consumers.

2. The computer system of claim 1, wherein receiving user interaction requesting a next best split comprises:

receiving information indicating the user clicking on a portion of the decision tree, the portion associated with the selected leaf node.

3. The computer system of claim 1, wherein identifying parameters usable to further segment consumers associated with the selected leaf node comprises:

identifying the parameters based on one or more split parameters identifying parameters of consumers to be optimized.

4. The computer system of claim 3, wherein the one or more parameters to be optimized include a variable to be optimized, and wherein the statistical parameters are calculated based on the variable to be optimized.

5. The computer system of claim 1, wherein providing an indication of the identified plurality of parameters as possible split decision nodes comprises dynamically displaying each possible split decision node and associated statistical information.

6. The computer system of claim 1, wherein the operations further comprise:

maintaining monitoring data, for later review by users, describing each possible split decision node including leaf nodes associated with the possible split decision node and statistical parameters for the possible split decision node.

7. A computer-implemented method comprising:

obtaining information describing process steps included in a user-defined process flow, with one or more of the process steps being associated with a corresponding one or more decision trees usable to determine information associated with a consumer based on inputs describing parameters of the consumer;

providing, for presentation to a user, a graphical user interface depicting a decision tree comprising;

a plurality of nodes including a plurality of decision nodes and a plurality of leaf nodes;

wherein each decision node is associated with one or more leaf nodes, and wherein and each of the leaf nodes is associated with different value ranges of a parameter of consumers associated with the particular decision node;

receiving, from a user, user interaction requesting a next best split of a selected leaf node associated with a selected consumer parameter;

in response to the next best split request;

automatically, by the computing device, identifying a plurality of parameters of consumers usable to further segment consumers associated with the selected leaf node;

identifying, for each of the identified plurality of parameters, one or more leaf nodes associated with different value ranges of the respective parameter;

providing, for presentation to the user, an indication of the identified plurality of parameters as possible split decision nodes, including statistical parameters indicating respective rankings of the possible split decision nodes, wherein the rankings of the possible split decision nodes are determined automatically by the computing device;

receiving user input selecting a first of the possible split decision nodes to be merged into the decision tree;

incorporating, into the decision tree, the selected first possible decision node and the identified one or more leaf nodes associated with the selected first possible decision node

receiving user input associated with calculating proposed outcomes for association with the leaf nodes, the user input indicating constraints to be optimized; and

determining, based on the constraints, associations between proposed outcomes and leaf nodes,

wherein the updated decision tree is usable to segment consumers and determine associated outcomes for the consumers.

8. The computer-implemented method of claim 7, wherein receiving user interaction requesting a next best split comprises:

receiving information indicating the user clicking on a portion of the decision tree, the portion associated with the selected leaf node.

9. The computer-implemented method of claim 7, wherein identifying parameters usable to further segment consumers associated with the selected leaf node comprises:

identifying the parameters based on one or more split parameters identifying parameters of consumers to be optimized.

10. The computer-implemented method of claim 9, wherein the one or more parameters to be optimized include a variable to be optimized, and wherein the statistical parameters are calculated based on the variable to be optimized.

11. The computer-implemented method of claim 7, wherein providing an indication of the identified plurality of parameters as possible split decision nodes comprises dynamically displaying each possible split decision node and associated statistical information.

12. The computer-implemented method of claim 7, further comprising:

maintaining monitoring data, for later review by users, describing each possible split decision node including leaf nodes associated with the possible split decision node and statistical parameters for the possible split decision node.