Feasibility of Automated Estimation of Software Development Effort in Agile Environments
Alhad Vinayak Sapre · OhioLink ETD Center (Ohio Library and Information Network) · 2012
Software development effort estimation is the process of predicting the most realistic use of effort required to develop or maintain software based on incomplete, uncertain and/or noisy input.Poor estimation may be the cause of significant challenges in project management and in the software quality.Therefore, it is important to make the use of estimation models and appropriate techniques to avoid losses caused by poor estimation.To address the estimation related issues in case of agile software development, we are proposing an automated estimation approach "Auto-Estimate", which can be a complement for Planning Poker, a time-consuming method, used widely in agile environments.Auto-Estimate is a machine learning-based estimation method for agile effort prediction, which makes use of predictors extracted from story cards (a prominent way to specify requirements in agile projects).Auto-Estimate can learn the estimation process during the initial stages of a project and be used as a tool for verification of Planning Poker estimates later in the project.It can fit naturally into agile development practices, as its granularity can be adjusted to a single story card level, as well as an iteration and a whole project.We have analyzed this estimation approach on aspects such as accuracy, applicability and value and presented the results.iii Dedication To my parents Chapter 1: Introduction In this section, I have stated the results of our experiments and analyzed them from different perspectives such as cost, accuracy, distribution of estimates and statistical measures.Next, I have presented my views about whether Planning Poker should be eliminated as a tool for estimation or should we use Auto-Estimate as a verification tool for estimates of Planning Poker.Finally, I have put forward the conclusions of the analysis using Auto-Estimate.