Collaborative decision engine for quality function deployment

• US 10,438,143 B2
• Filed: 09/28/2015
• Issued: 10/08/2019
• Est. Priority Date: 09/28/2015
• Status: Active Grant

• Current Assignee: Bank of America Corp.
• Original Assignee: Bank of America Corp.
• Inventors: Vemula, Srikanth, Gaddam, Sunil Reddy, Purushothaman, Sasidhar
• Primary Examiner(s): Shmatov, Alexey
• Assistant Examiner(s): Alabi, Oluwatosin O
• Application Number:
  US14/867,244
• Publication Number:
  US 20170091633A1
• Time in Patent Office: 1,471 Days
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling

Abstract

Disclosed is systems, methods, and computer program products that provide for a technique for reducing computing resources, storage space needs, and network bandwidth associated with collaborative decision making. More particularly, this disclosure relates to a system for performing automatic predictive decision making using predictive fit models derived from previous user responses and the user characteristics of each responding user, and using the results to reduce the amount of computing and operational resources needed to operate a collaborative decision engine.

Claims

1. A system for reducing computing resources associated with collaborative decision-making, said system comprising:

one or more memory devices having computer readable program code stored thereon; and

one or more processing devices operatively coupled to the one or more memory devices, wherein the one or more processing devices are configured to execute the computer readable program code to;

identify a plurality of users that use a feature of an application, wherein the application is a virtual or physical process or task associated with an entity, and wherein the feature is at least a component of the application;

retrieve, from a database, sets of user characteristics for each of the plurality of users;

identify criteria associated with the feature, wherein the criteria are aspects of the feature that affect how the plurality of users perceives the feature or makes a decision regarding the feature;

determine predictive responses for each criteria of the feature based on stored prediction fit models for each criteria and each set of user characteristics represented in the identified plurality of users, wherein the stored predictive fit models provide a predictive analysis of how a future user would respond to a criteria question based on a user characteristic, and wherein the stored predictive fit models are determined by analyzing historical information comprising;

historical feature surveys previously provided to previous users for the feature, wherein the historical feature surveys included a request for previous user input for each criteria of the feature and a previous user weightings of each criteria of the feature;

received previous user input for each criteria of the feature and the previous user weightings of each criteria of the feature;

determined previous user group weightings for each previous user input for each criteria based on a user group associated with each previous user; and

determined criteria response distributions for characteristics of each previous user based on the previous user input, the previous user weightings of each criteria of the feature; and

the determined previous user group weightings;

determine confidence scores for each predictive response associated with a probability of accuracy of each predictive response based on the predictive response, the stored fit models for each criteria, and the sets of user characteristics;

determine that a first predictive response from the predictive responses for each criteria does not meet a threshold confidence score;

in response to determining that the first predictive response does not meet the threshold confidence score, transmit a new survey to computing devices associated with the plurality of users comprising at least one question related to the predictive response that does not meet the threshold confidence score;

receive, from the computing devices associated with the plurality of users, an actual response as user input associated with the new survey from the plurality of users;

replace the first predictive response with the actual response; and

display the actual response, along with predictive responses that do meet respective determined threshold confidence scores, on a computing device user interface associated with a manager of the application.

2. The system of claim 1, wherein the sets of user characteristics are characteristics of each responding user and include at least one of user group, geographical location, time zone, work experience, and technical background.

3. The system of claim 1, wherein the criteria associated with the feature include at least one of ease of use, intuitiveness, speed, maintainability, effectiveness, amount of support required, usefulness, whether the feature should be retained, and an extent to which a responding user utilizes the feature.

4. The system of claim 1, wherein determining the confidence score further comprises:

determining a goodness of fit for the predictive fit model based on a correlation between the predictive fit model and its respective criteria distribution; and

calculating a confidence score based on the goodness of fit.

5. The system of claim 1, wherein the criteria response distributions for characteristics of each previous user is analyzed by a pattern analyzer that determines a linear or nonlinear regression fit model for the criteria response distribution.

6. A computer program product for reducing computing resources associated with collaborative decision-making, the computer program product comprising a non-transitory computer readable medium comprising computer readable instructions, the instructions comprising instructions for:

identifying a plurality of users that use a feature of an application, wherein the application is a virtual or physical process or task associated with an entity, and wherein the feature is at least a component of the application;

retrieving, from a database, sets of user characteristics for each of the plurality of users;

identifying criteria associated with the feature, wherein the criteria are aspects of the feature that affect how the plurality of users perceives the feature or makes a decision regarding the feature;

determining predictive responses for each criteria of the feature based on stored prediction fit models for each criteria and each set of user characteristics represented in the identified plurality of users, wherein the stored predictive fit models provide a predictive analysis of how a future user would respond to a criteria question based on a user characteristic, and wherein the stored predictive fit models are determined by analyzing historical information comprising;

historical feature surveys previously provided to previous users for the feature, wherein the historical feature surveys included a request for previous user input for each criteria of the feature and a previous user weightings of each criteria of the feature;

received previous user input for each criteria of the feature and the previous user weightings of each criteria of the feature;

determined previous user group weightings for each previous user input for each criteria based on a user group associated with each previous user; and

determined criteria response distributions for characteristics of each previous user based on the previous user input, the previous user weightings of each criteria of the feature; and

the determined previous user group weightings;

determining confidence scores for each predictive response associated with a probability of accuracy of each predictive response based on the predictive response, the stored fit models for each criteria, and the sets of user characteristics;

determining that a first predictive response from the predictive responses for each criteria does not meet a threshold confidence score;

in response to determining that the first predictive response does not meet the threshold confidence score, transmitting a new survey to computing devices associated with the plurality of users comprising at least one question related to the predictive response that does not meet the threshold confidence score;

receiving, from the computing devices associated with the plurality of users, an actual response as user input associated with the new survey from the plurality of users;

replacing the first predictive response with the actual response; and

displaying the actual response, along with predictive responses that do meet respective determined threshold confidence scores, on a computing device user interface associated with a manager of the application.

7. The computer program product of claim 6, wherein the sets of user characteristics are characteristics of each responding user and include at least one of user group, geographical location, time zone, work experience, and technical background.

8. The computer program product of claim 6, wherein the criteria associated with the feature include at least one of ease of use, intuitiveness, speed, maintainability, effectiveness, amount of support required, usefulness, whether the feature should be retained, and an extent to which a responding user utilizes the feature.

9. The computer program product of claim 6, wherein determining the confidence score further comprises:

determining a goodness of fit for the predictive fit model based on a correlation between the predictive fit model and its respective criteria distribution; and

calculating a confidence score based on the goodness of fit.

10. The computer program product of claim 6, wherein the criteria response distributions for characteristics of each user is analyzed by a pattern analyzer that determines a linear or nonlinear regression fit model for the criteria response distribution.

11. A computer implemented method for reducing computing resources associated with collaborative decision-making, said computer implemented method comprising:

identifying, via a processing device, a plurality of users that use a feature of an application, wherein the application is a virtual or physical process or task associated with an entity, and wherein the feature is at least a component of an application;

retrieving, via a processing device, from a database, sets of user characteristics for each of the plurality of users;

identifying, via a processing device, criteria associated with the feature, wherein the criteria are aspects of the feature that affect how the plurality of users perceives the feature or makes a decision regarding the feature;

determining, via a processing device, predictive responses for each criteria of the feature based on stored prediction fit models for each criteria and each set of user characteristics represented in the identified plurality of users, wherein the stored predictive fit models provide a predictive analysis of how a future user would respond to a criteria question based on a user characteristic, and wherein the stored predictive fit models are determined by analyzing historical information comprising;

historical feature surveys previously provided to previous users for the feature, wherein the historical feature surveys included a request for previous user input for each criteria of the feature and a previous user weightings of each criteria of the feature;

received previous user input for each criteria of the feature and the previous user weightings of each criteria of the feature;

determined previous user group weightings for each previous user input for each criteria based on a user group associated with each previous user; and

determined criteria response distributions for characteristics of each previous user based on the previous user input, the previous user weightings of each criteria of the feature; and

the determined previous user group weightings;

determining, via a processing device, confidence scores for each predictive response associated with a probability of accuracy for each predictive response based on the predictive response, the stored fit models for each criteria, and the sets of user characteristics;

determining, via a processing device, that a first predictive response from the predictive responses for each criteria does not meet a threshold confidence score;

in response to determining that the first predictive response does not meet the threshold confidence score, transmitting, via a processing device, a new survey to the plurality of users comprising at least one question related to the predictive response that does not meet the threshold confidence score;

receiving, via a processing device, from the computing devices associated with the plurality of users, an actual response as user input associated with the new survey from the plurality of users;

replacing, via a processing device, the first predictive response with the actual response; and

displaying, via a processing device, the actual response, along with predictive responses that do meet respective determined threshold confidence scores, on a computing device user interface associated with a manager of the application.

12. The computer implemented method of claim 11, wherein the sets of user characteristics are characteristics of each responding user and include at least one of user group, geographical location, time zone, work experience, and technical background.

13. The computer implemented method of claim 11, wherein the criteria associated with the feature include at least one of ease of use, intuitiveness, speed, maintainability, effectiveness, amount of support required, usefulness, whether the feature should be retained, and an extent to which a responding user utilizes the feature.

14. The computer implemented method of claim 11, wherein determining the confidence score further comprises:

determining a goodness of fit for the predictive fit model based on a correlation between the predictive fit model and its respective criteria distribution; and

calculating a confidence score based on the goodness of fit.

15. The computer implemented method of claim 11, wherein the criteria response distributions for characteristics of each previous user is analyzed by a pattern analyzer that determines a linear or nonlinear regression fit model for the criteria response distribution.