Feature Estimation Model Using Epics, Sprints, and Historical Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing software development processes face inefficiencies in feature estimation, particularly in agile methodologies, due to the time-consuming and often inaccurate nature of estimating work items using user stories and story points, leading to potential delays and inaccuracies in project timelines.
Innovation Solution
A large-scale feature estimation model that automatically decomposes features into epics and sprints, adjusts sprint counts based on historical data, and generates estimates without relying on user stories, using a machine learning system to control software development processes, including CI/CD systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual estimation using user stories and story points is used, then estimation flexibility is maintained, but estimation time consumption increases and accuracy decreases
Solution Approach 1:
The patent replaces manual estimation mechanisms with an automated machine learning model. The system automatically decomposes features into epics and sprints, calculates sprint counts based on historical data, and generates estimates without human intervention, thereby eliminating time consumption while maintaining or improving accuracy through data-driven predictions
Solution Approach 2:
The patent changes the estimation parameters from traditional user story-level details to epic and sprint-level aggregates. By operating at higher levels of the requirements hierarchy (features → epics → sprints), the system reduces the granularity of input data while maintaining estimation validity, thus reducing time requirements without sacrificing measurement precision
2Measurement precision
If detailed user stories and story points are used for estimation, then estimation completeness is improved, but process complexity increases
Solution Approach 1:
The patent segments the estimation process into distinct hierarchical levels: feature decomposition into epics, epic decomposition into sprints, and sprint estimation. This segmentation allows the system to handle complexity at appropriate granularities, estimating sprints based on historical data patterns rather than requiring detailed analysis of every user story
Solution Approach 2:
The patent introduces epic and sprint-level aggregates as intermediary structures between features and user stories. These intermediaries serve as the actual estimation units, absorbing the complexity of underlying user stories while presenting a simplified estimation interface at the epic/sprint level, thus reducing overall process complexity while maintaining completeness
3Productivity
If automated estimation model is implemented, then estimation speed increases, but model complexity and data requirements increase
Solution Approach 1:
The patent performs preliminary decomposition of features into epics and sprints before estimation occurs. Historical data is pre-processed and stored in the requirements hierarchy structure, allowing the estimation model to directly query pre-aggregated sprint counts from history rather than performing complex real-time analysis, thus reducing model complexity while maintaining high estimation speed
Solution Approach 2:
The patent designs the estimation model to serve multiple functions simultaneously: it decomposes features into epics, determines sprint counts, adjusts for historical patterns, and generates estimates. By making the model multi-functional and leveraging existing CI/CD system data structures, the patent reduces the need for separate specialized components, thereby limiting the increase in overall system complexity despite automation
Data Source
AI summary
A method in an illustrative embodiment comprises configuring a software development process to include a plurality of features in respective distinct feature domains, and for each of the features, decomposing the feature into a plurality of epics, determining a number of sprints associated with each of the epics of the feature, determining a total number of sprints across all of the epics of the feature, adjusting the total number of sprints based at least in part on historical data, and automatically generating an estimate for completion of the feature based at least in part on the adjusted total number of sprints. One or more characteristics of the software development process are controlled based at least in part on the estimates generated for respective ones of the features. The software development process in some embodiments is illustratively an agile software development process implemented in a continuous integration/continuous deployment (CI/CD) system.