Assisted analytics
AssigneeDimensional Insight Incorporated
InventorsFrederick A Powers, James Clark
Life of this patent
Estimating the term…
Abstract
Predicting a change in a business workflow performance measure includes a measure factory rule produced outlier detecting step, a difference-over-time outlier contributing data element determining step, a different business workflow performance measure identifying step for the contributing data element, an outlier prediction step for the different business workflow performance measure as derived from the contributing data element, and a step for alerting a user when the predicted outlier falls outside a pattern for business workflow performance measures.
Claims18
A method of predicting a change in a business workflow performance measure, the method comprising:
- detecting an outlier in a first business workflow performance measure output value, wherein the outlier in the measure output value is produced by applying at least one factory rule to data sourced from business workflow data corresponding to a business workflow;
- determining a contributing data element used by the at least one factory rule to produce the outlier in the measure output value, the determining based on a difference of values of the contributing data element over time;
- identifying a second business workflow performance measure that is configured to use the contributing data element, the second business workflow performance measure identified from a plurality of business workflow performance measures that includes the first business workflow performance measure, wherein the second business workflow performance measure is a different measure than the first business workflow performance measure;
- automatically predicting an outlier in a measure output value for the second business workflow performance measure based on the determined contributing data element;
- and
- alerting a user of the predicted outlier when the predicted outlier falls outside a pattern for business workflow performance measures of the business workflow.
A method of predicting a change in a business workflow performance measure, the method comprising:
- detecting an outlier in a first business workflow performance measure output value wherein the outlier is produced by applying at least one factory rule to data sourced from business workflow data corresponding to a business workflow;
- identifying a second business workflow performance measure that is configured to use the first business workflow performance measure output value, wherein the second business workflow performance measure is identified based on an impact of the first business workflow performance output value on the second business workflow performance measure;
- automatically predicting a change in outcome from performing the second business workflow performance measure based on use of the first business workflow performance measure output value by the second business workflow performance measure;
- and
- alerting a user of the predicted change in outcome when the predicted change in outcome is outside of a pattern of outcomes for business workflow performance measures of the business workflow.
A method of predicting an outlier outcome of a set of outcomes produced by a measure factory, the method comprising:
- applying a first factory rule and a source rule to data sourced from a data set for which the source rule applies to generate a first outcome;
- determining that the first outcome comprises a value that is outside of a normalization of the first outcome;
- monitoring factory rule data usage for a plurality of factory rules, other than the first factory rule, for use of at least one data entry in a data set that contributed to the first outcome, wherein the at least one data entry comprises a value that exceeds a threshold for values of the at least one data entry;
- and
- predicting that at least one outcome generated by applying at least one of the plurality of factory rules, other than the first factory rule will include an outlier value that is outside of a normalization of the at least one outcome when the at least one data entry is used to produce the at least one outcome.
Dependencies are inferred from the claim text.