Practical Software Project ImprovementsUsing actionable predictive models and solutions<br />Panel Members:<br />Jacky Keu...
Welcome<br />
Today’s Overview <br />
Introduction<br />
Can we use our metrics to change projects?<br />
Actionable Metrics Research<br />For example..<br />Programmer actions vs. software defects<br />Some actions introduce de...
Elementary programmer actions<br />Opening files<br />Writing tests<br />Running programs<br />…etc<br />
Defects and Programmer Actions<br />Can we correlate defects with programmer actions (failure-correlated actions)? <br />C...
Solutions need to be practical<br />The truth is not so simple …<br />Cannot easily change standard practices <br />Change...
The story               <br />        begins…<br />
Failure is a Four-Letter Word: A Satire in Empirical Research<br />Andreas Zeller, Thomas Zimmermann, Christian Bird<br />
Panelist Discussion<br />Jacky Keung, Wang Qing, Martin Shepperd, Emila Mendes<br />
Practical Software Project Improvements using Actionable Predictive Models and SolutionsMartin ShepperdBrunel Uni, UK<br /...
Casuality requires …<br />X covaries with Y<br />X has temporal precedence<br />No more plausible competing explanations<b...
The Machine Learnerotron!<br />15<br />Martin Shepperd<br />numbers<br />answers<br />
Can we use our metrics to change projects?<br />
Your views about empirical results?<br />How do you use them?<br />Do you change your SE processes based on these results?...
Discussions<br />Correlations do not imply causations<br />Do not confuse causes and symptoms<br />Generalization, samples...
How to make findings actionable?<br />An empirical finding is more valuable the more actionable it is…<br />What is the co...
Questions?<br />
Upcoming SlideShare
Loading in …5
×

Promise 2011: Panel - "Practical Software Project Improvements using Actionable Predictive Models and Solutions"

3,011 views
2,964 views

Published on

Promise 2011:
Panel - "Practical Software Project Improvements using Actionable Predictive Models and Solutions"
Jacky Keung.

Published in: Technology, Education
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
3,011
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide
  • This template can be used as a starter file for presenting training materials in a group setting.SectionsSections can help to organize your slides or facilitate collaboration between multiple authors. On the Home tab under Slides, click Section, and then click Add Section.NotesUse the Notes pane for delivery notes or to provide additional details for the audience. You can see these notes in Presenter View during your presentation. Keep in mind the font size (important for accessibility, visibility, videotaping, and online production)Coordinated colors Pay particular attention to the graphs, charts, and text boxes.Consider that attendees will print in black and white or grayscale. Run a test print to make sure your colors work when printed in pure black and white and grayscale.Graphics, tables, and graphsKeep it simple: If possible, use consistent, non-distracting styles and colors.Label all graphs and tables.
  • This is another option for an overview slide.
  • This is another option for an overview using transitions to advance through several slides.
  • Use a section header for each of the topics, so there is a clear transition to the audience.
  • Use a section header for each of the topics, so there is a clear transition to the audience.
  • What will the audience be able to do after this training is complete?Briefly describe each objective and how the audience will benefit from this presentation.
  • What will the audience be able to do after this training is complete?Briefly describe each objective and how the audience will benefit from this presentation.
  • Promise 2011: Panel - "Practical Software Project Improvements using Actionable Predictive Models and Solutions"

    1. 1. Practical Software Project ImprovementsUsing actionable predictive models and solutions<br />Panel Members:<br />Jacky Keung (HK Polytechnic University)<br />Wang Qing (Chinese Academy of Science)<br />Martin Shepperd(Brunel University)<br />Emila Mendes (Auckland University)<br />
    2. 2. Welcome<br />
    3. 3. Today’s Overview <br />
    4. 4. Introduction<br />
    5. 5. Can we use our metrics to change projects?<br />
    6. 6. Actionable Metrics Research<br />For example..<br />Programmer actions vs. software defects<br />Some actions introduce defects<br />Measuring likelihood of introducing defects<br />Early warning system for programmers<br />
    7. 7. Elementary programmer actions<br />Opening files<br />Writing tests<br />Running programs<br />…etc<br />
    8. 8. Defects and Programmer Actions<br />Can we correlate defects with programmer actions (failure-correlated actions)? <br />Can we isolate defect-related actions?<br />Can we prevent defects by preventing actions?<br />If all yes, we can build a model and to prevent defects!! <br />
    9. 9. Solutions need to be practical<br />The truth is not so simple …<br />Cannot easily change standard practices <br />Change a process leads to other concerns<br />Productivity, process, outcomes <br />Is that what we have been doing over the past 30 years? <br />…<br />
    10. 10. The story <br /> begins…<br />
    11. 11. Failure is a Four-Letter Word: A Satire in Empirical Research<br />Andreas Zeller, Thomas Zimmermann, Christian Bird<br />
    12. 12. Panelist Discussion<br />Jacky Keung, Wang Qing, Martin Shepperd, Emila Mendes<br />
    13. 13. Practical Software Project Improvements using Actionable Predictive Models and SolutionsMartin ShepperdBrunel Uni, UK<br />Actionable -> causality<br />Machine learning ≠ magic<br />13<br />Martin Shepperd<br />
    14. 14. Casuality requires …<br />X covaries with Y<br />X has temporal precedence<br />No more plausible competing explanations<br />✓<br />✓<br />✗<br />14<br />Martin Shepperd<br />
    15. 15. The Machine Learnerotron!<br />15<br />Martin Shepperd<br />numbers<br />answers<br />
    16. 16. Can we use our metrics to change projects?<br />
    17. 17. Your views about empirical results?<br />How do you use them?<br />Do you change your SE processes based on these results?<br />What are the issues? <br />Share your views…<br />
    18. 18. Discussions<br />Correlations do not imply causations<br />Do not confuse causes and symptoms<br />Generalization, samples<br />Cherry-picking, deliberately suppressing other results<br />Beware of fraud<br />Threats to validity <br />Careful use of machine learning (careful comparison with the state of the art)<br />
    19. 19. How to make findings actionable?<br />An empirical finding is more valuable the more actionable it is…<br />What is the consequence of the result?<br />Should I change things?<br />How?<br />What is the risk of this change?<br />Result should provide for its potential implications.<br />Immediately useful?<br />Requires changes in other aspects?<br />Risk of change?<br />
    20. 20. Questions?<br />

    ×