SOCIAL METRICS INCLUDED IN PREDICTION MODELS ON SOFTWARE ENGINEERING: A MAPPING STUDY

SOCIAL METRICS INCLUDED IN PREDICTION

MODELS ON SOFTWARE ENGINEERING: A MAPPING

STUDY

Igor Wiese, Filipe Côgo, Reginaldo Ré, Igor Steinmacher

and Marco Aurélio Gerosa

2

MOTIVATION SCENARIO

Subsystem A

Class

aClass

c

Class

b

Subsystem B

Class

d

Class

e

Software systems are

composed byArtifacts that

dependes oneeach other

HISTORICAL

SIDE

3

PROBLEM STATEMENT

Even when social metrics were

considered, they were classified as

part of other dimensions, such as

process, history, or change.

Is not clear yet which social metrics

are used in prediction models and

what are the results of their use in

different contexts.

4

OBJECTIVE

Classify

Identify

5

RESEARCH QUESTION

Which social metrics were used in prediction

models?

Did the social metrics have positive effect

when they were considered as predictor?

RQ1

RQ2

We found that previous SLR did not discussed explicitly about

social metrics

inconsistent terminology for classifying social metrics and often do

not report their individual result

we identified papers describing evidences about the effects of social

metrics

we summarized the proposed classification, linking each group of

metrics to the applicability of prediction models

we mapped in which application each group of social metrics were

used so far

6

SNOWBALLING PROCESS METHOD

Papers that explicitly mention only code metrics,

objected oriented metrics, or static metrics

7

DATA EXTRACTION

ICSE

10 papers

FSE

6 papers

PROMISE

5 papers

48 Primary Papers, 103 distinct authors

8

RQ1: WHICH SOCIAL METRICS WERE

USED IN PREDICTION MODELS?

9

RQ1: WHICH SOCIAL METRICS WERE

USED IN PREDICTION MODELS?

Group (3)

Category (9)

Sub-category (51)

10

RQ1: WHICH SOCIAL METRICS WERE

USED IN PREDICTION MODELS?

11

RQ1: WHICH SOCIAL METRICS WERE

USED IN PREDICTION MODELS?

12

RQ2: DID THE SOCIAL METRICS HAVE

POSITIVE EFFECT WHEN THEY WERE

CONSIDERED AS PREDICTOR?

6 papers reported

negative effects

21 papers reported

positive effects

2 papers reported

neutral effects

13

RQ2: DID THE SOCIAL METRICS HAVE

POSITIVE EFFECT WHEN THEY WERE

CONSIDERED AS PREDICTOR?

CONCLUSIONS

• social metrics were classified as part of other dimension, such

as process, history, or change

• Considering the results published so far, it could be risky to

draw generalized conclusions about social metrics.

• New opportunities of research concerning social metrics

• different techniques and limited number of software

projects in different contexts.

• To consider large scale and longitudinal analysis

• To investigate the effectiveness of social metrics to build

prediction models

14

OUR RESEARCH

15

Artifacts a1

Artifacts a2

time

Change coupling

commit

A change dependency indicates that two

artifacts changed together (co-changed)

in the past, making them evolutionarily connected

16

PROBLEM STATEMENT

File a1

File a2

time

Change coupling

commit

+ SOCIAL

+ HISTORICALD´ambros - benchmark

Tracy hall - SLR, etc

D´ambros – OSS

Kirbas/Ayse Bener – Industrial

Gustavo Oliva,

Markus Geipel

http://lapessc.ime.usp.br/

Thank you

[email protected]