Techniques for improving on-time delivery in large software development organizations

No Thumbnail Available

URL

Journal Title

Journal ISSN

Volume Title

Sähkötekniikan korkeakoulu | Master's thesis

Date

2019-03-11

Department

Major/Subject

Software and Service Engineering

Mcode

SCI3069

Degree programme

CCIS - Master’s Programme in Computer, Communication and Information Sciences (TS2013)

Language

en

Pages

70+5

Series

Abstract

The following work investigated analysis techniques applicable to software developer conversations, collected from an online collaboration platform, with the end goal of improving on-time delivery of software products. Specifically, the possible associations between discussions and delays in the planning and implementation process were investigated. In addressing this research problem, the conversations were abstracted and modeled as time series in two distinct approaches. Firstly, focusing on discussion structure, an automatic text classifier was used to reduce every comment to a category best describing its intent, such as asking or sharing information. Secondly, looking into content evolution, a thematic analysis was employed with codes representing discussion topic, such as technical discussion. Time series clustering was then employed to identify the most prevalent patterns of progression and investigate their association to delayed deliveries. Although results showed no clear link between discussion structure and delivery, not one pattern being clearly indicative of a delay, patterns in topic evolution led to the development of themes characterizing delays, providing their probable causes, and the formulation of recommendations for software engineering practitioners. For the analyzed project, the most notable causes of delay were prolonged discussions during planning, ambiguous technical specifications or additional work required close to the delivery date. This case-specific insight is critical to performing accurate assessments and creating customized action plans for improvement.

Description

Supervisor

Lassenius, Casper

Thesis advisor

Damian, Daniela

Keywords

software engineering, conversation analysis, time-series clustering, text mining, thematic analysis

Other note

Citation