metrics - you can't control the unfamiliar
DESCRIPTION
Paper: You Can't Control the Unfamiliar: A Study on the Relations Between Aggregation Techniques for Software Metrics Authors: Bogdan Vasilescu, Alexander Serebrenik and Mark Van Den Brand Session: Research Track 11 - MetricsTRANSCRIPT
/ W&I / MDSE PAGE 0 5-10-2011
Metrics are usually computed at a low level:
classes, methods, …
Multitude of data values obscures a general
picture of the system maintainability
/W&I / MDSE PAGE 1 5-10-2011
That we are actually interested in!
/W&I / MDSE PAGE 2 5-10-2011
You Can't Control the Unfamiliar:
A Study on the Relations
Between Aggregation
Techniques for Software Metrics
Bogdan Vasilescu
Alexander Serebrenik
Mark van den Brand
Two kinds of aggregation
Same artifact, different
metrics
Same metrics, different
artifacts
/W&I / MDSE PAGE 4 5-10-2011
Various techniques can be
found in the literature
Same metrics, different
artifacts
/W&I / MDSE PAGE 5 5-10-2011
Traditional: mean,
median, sum, …
Econometric
inequality indices:
Gini, Theil, Hoover,
Kolm, Atkinson
Various techniques can be
found in the literature
Same metrics, different
artifacts
/W&I / MDSE PAGE 6 5-10-2011
Traditional: mean,
median, sum, …
Econometric
inequality indices:
Gini, Theil, Hoover,
Kolm, Atkinson
Which
aggregation
technique
should we
use?
Questions
1. Which and to what extent do the different
aggregation techniques agree?
2. What is the nature of the relation between the
various aggregation techniques?
3. How does the correlation coefficient change as the
systems evolve?
/W&I / MDSE PAGE 7 5-10-2011
Qualitas Corpus 20101126
/W&I / MDSE PAGE 8 5-10-2011
• Qualitas Corpus 20101126r, 106 systems
• FitJava v1.1, 2 packages, 2240 SLOC
• NetBeans v6.9.1, 3373 packages 1890536 SLOC.
1) Agreement between diff techniques
• Agreement:
• Aggregation: Class SLOC Package
• Techniques agree if they rank the packages similarly
/W&I / MDSE PAGE 9 5-10-2011
We use rank-based correlation coefficient: Kendall’s
1) Agreement: different inequality indices?
• Gini, Theil, Hoover, Atkinson – agree
• aggregates obtained convey the same information
• Kolm does not!
/W&I / MDSE PAGE 10 5-10-2011
1) Agreement: traditional and ineq indices?
• mean
• Kolm: strong (0,8) and statistically significant (92%)
• median, standard deviation, and variance
• sum
• does not correlate with any other aggregation technique
/W&I / MDSE PAGE 11 5-10-2011
2) Nature of the relation: Typical patterns
• Theil is known to be more
sensitive to the rich
• Theil increases faster
when Gini increases
/W&I / MDSE PAGE 12 5-10-2011
• Linear relation with a “fat”
head
Which aggregation technique? (1)
• Theil, Hoover, Gini and Atkinson agree
• Any can be chosen from the correlation point of view
• Some might be “better” in each specific case
• easy to interpret: Gini [0,1]
• provide additional insights: Theil (explanation)
• negative values: Gini, Hoover
− affects the domain!
• sensitive for high values: Theil, Atkinson
• deviations from uniformity: Gini, Hoover
/ W&I / MDSE PAGE 13 5-10-2011
Which aggregation technique? (2)
• Kolm and mean agree
• Kolm is reliable for skewed distributions
− better alternative (“by no means”)
• Not in the paper:
− agreement observed for NOC
− but not for DIT!
/ W&I / MDSE PAGE 14 5-10-2011
Conclusions
/W&I / MDSE PAGE 15 5-10-2011