Since November 2003, Martin Shepperd and Carolyn Mair, in association with STTF and BT, have been working on a 3-year EPSRC funded project, MeLLow (meta level learning for software projects), at Brunel University.
The aim of MeLLow is to develop and apply a framework to systematically interpret and exploit the increasing number of empirical results derived from comparing software project effort prediction systems. The project has highlighted several areas of inconsistencies which make combining and comparing results problematic.
For example, we suggested why inconsistencies within and between results in studies on comparisons of prediction methods for software engineering project cost estimation might arise, and have proposed ways to reduce them in the future. We hope our suggestions and proposals might produce more readily useful data from which a body of knowledge, and ultimately theories, can be built.
In order to discuss and expand on these propositions with experts from around the world we our holding a workshop on 17th and 18th October 2006. Invited speakers are presenting work on a number of related issues. Please see the schedule for abstracts, slides and photos of the speakers.
Day 1: 17th October 2006
Decoding empirical data: what do we know about cost modelling?
|12:30||Registration and buffet lunch||SJ004|
|13:30||Introduction and Welcome||Martin Shepperd and Ken Darby-Dowman (Brunel, UK)||SJ004|
|13:45||IT Research – My Big Challenges (slides)||Dave Griffiths (BT, UK)||SJ004|
|14:15||Estimating Project Success (slides)||June Verner (NICTA, Australia)||SJ004|
|15:05||Practical aspects of data collection in industry (slides)||Pekka Forselius (STTF, Finland)||SJ004|
|15:45||How can you know what my data means?||Chris Lokan (UNSW@ADFA, Australia)||SJ004|
|16:35||Results from a systematic review ... (slides)||Emilia Mendes (U. of Auckland, NZ)||SJ004|
|17:15||Comparison of group estimation studies (slides)||Kjetil Moløkken-Østvold (Simula Labs, Norway)||SJ004|
|17:55||Summary||Martin Shepperd (Brunel, UK)||SJ004|
|18:00||Drinks and dinner||The Load of Hay pub|
Day 2: 18th October 2006
Activities: protocol? methods? quality? Uptake by researchers?
|09:30||Introduction + review of 17th October||Martin Shepperd (Brunel, UK)||SJ004|
|09:35||Meta-analysis issues (slides)||Carolyn Mair (Brunel, UK)||SJ004|
|10:05||How repeatable are systematic reviews? (slides)||Steve MacDonell (AUT, NZ)||SJ004|
|10:45||Using CVS repositories for software engineering estimation (slides)||Rahul Premraj (U of Saarbrucken, Germany)||SJ004|
|11:15||Evaluating data quality issues from an industrial data set (slides)||Gernot Liebchen and Bheki Twala (Brunel, UK), Mark Stephens (EDS, UK)||SJ004|
|12:00||Effort Prediction Models for Interval Estimation (slides)||Ioannis Stamelos and Lefteris Angelis (Aristotle University of Thessaloniki, Greece)||SJ004|
|12:30||On the difficulties in determining effort estimation accuracy in practice (slides)||Ursula Passing (Germany)||SJ004|
|14:00||Chris Lokan, Steve MacDonell, Mark Stephens, June Verner and Mark Harman (chair)||Reverse panel||SJ004|
|14:45||All (in groups)||Practical sessions||SJ004|
|16:00||All||Group findings from practical sessions (slides)||SJ004|
|17:00||Martin Shepperd (Brunel, UK)||Summary||SJ004|
|17:15||Farewell buffet and wine|
Finally, courtesy of Mark Harman, a meta-level limerick!
A meta-analyst from Brunel
Wanted a systematic review as well
In defining these terms
Opening a big can of worms
We descended the pathway to hell.