Software Analytics

In my PhD thesis I argued that we need to make our software analysis tools work beyond individual software systems at the level of software ecosystems.

I gave two reasons for this: (1) by analyzing a system in context we can better understand it and (2) analyzing software ecosystems in themselves can provide insight that otherwise would not be possible.

Machine Learning approaches are good in solving problems that have less information. In most cases, the software domain problems characterize as a process of learning that depend on the various circumstances and changes accordingly. A predictive model is constructed by using machine learning approaches and classified them into defective and non-defective modules. Machine learning techniques help developers to retrieve useful information after the classification and enable them to analyse data from different
perspectives. Machine learning techniques are proven to be useful in terms of software bug prediction. This
study used public available data sets of software modules and provides comparative performance analysis of different machine learning techniques for software bug prediction. Results showed most of the machine learning methods performed well on software bug datasets.

Nearly every large organization on Earth is involved in software development at
some level. Some organizations specialize in software development while other
organizations only participate in software development out of necessity. In both cases, the
performance of the software development matters. Organizations collect vast amounts of
data relating to software development. What do the organizations do with that data? That
is the problem. Many organizations fail to do anything meaningful with the data.

Another problem is knowing what data to collect. There are many options, but
certain data is more important than others. What data should a software development
organization collect?

This paper plans to answer that question and present a framework to gather the
right information and provide a score for an organization that produces software. The
score is not to be comparative between organizations, but to be comparative for a specific
organization over time.

The primary goal of this work is to provide a general framework for what a
software development organization should measure and how to report on those
measurements. The focus is providing a single number to represent the entire organization
and not just the development efforts. That single number is considered the Cumulative
Result Indicator (CRI) score. The secondary goal of this work is to provide a framework
for storing the necessary data.

Abstract—This position paper argues that fault classification
provides vital information for software analytics, and that
machine learning techniques such as clustering can be applied to
learn a project- (or organization-) specific fault taxonomy.
Anecdotal evidence of this position is presented as well as
possible areas of research for moving toward the posited goal.

Visual analytics (VA) is an emerging science at the crossroads of data and information visualization, graphics, data mining, and knowledge representation, with many successful applications in engineering, business and finance, security, geosciences, and e-governance and health. Tools using visualization, data mining, and data analysis are also prominently present in a different field: software maintenance. However, an integrated VA is relatively new for this field. In this paper, we discuss the specific challenges and particularities of applying VA in software engineering, highlight the added value of a VA approach, as distilled by us from several large-scale software engineering industrial projects.