Research Report: Explainable Software Analytics and Neural Networks

This report provides a comprehensive analysis of a research paper titled "Explainable Software Analytics" presented by Hoa, Truyen, and Aditya in 2018, focusing on the application of artificial neural networks in software engineering. The report explores the paper's intention, which is to evaluate software analytics, particularly explainable models, and their role in improving risk prediction and data management within organizations. It details the research methods employed, including an inductive approach and qualitative design, alongside data collection via secondary methods and content analysis for data analysis. The report highlights the problems addressed by the writers, such as the reluctance to trust predictions from analytics without understanding their rationale, and the results, which demonstrate the effectiveness of software analytics in enhancing risk prediction and overall performance. The findings emphasize the importance of explainability in software analytics models. The report concludes by summarizing the benefits of implementing software analytic models for accurate predictions and efficient risk management, advocating for the adoption of explainable software analytics to improve the understanding of machine predictions and protect data from potential risks. The paper's focus aligns with the assignment's objective to reflect on project practices and consider various factors in software engineering methodologies.

RESEARCH PAPER 0
Article Name: H.K., Dam, T. Tran and, A., Ghose, “Explainable software analytics,” In Proceedings of the
40th International Conference on Software Engineering: New Ideas and Emerging Results, vol. 12, no. 6,
pp. 53-56, 2018.
Artificial Neural Network

RESEARCH PAPER 1
Table of Contents
Introduction.................................................................................................................................................2
The intention of the paper............................................................................................................................2
Research methods........................................................................................................................................3
Research approach...................................................................................................................................3
Research design.......................................................................................................................................3
Data collection methods..........................................................................................................................4
Data analysis...........................................................................................................................................4
Problems highlighted by the writers............................................................................................................4
Results and findings....................................................................................................................................5
Conclusions of the paper.............................................................................................................................5
Conclusion...................................................................................................................................................5
References...................................................................................................................................................7

RESEARCH PAPER 2
Introduction
Artificial neural networks are defined as the computing networks which are inspired by the
biological neural networks that establish animal intelligence. It is based on the combination of
linked nodes which are defined as artificial neurons that convert models in biological
intelligence. Software analytics is a part of artificial intelligence that helps companies to analyze
larger data sets and enhance the quality of software and services effectively. The aim of the
investigation is to evaluate the concept behind software analytics and review a journal article
based on artificial neural networks. The chosen paper is “Explainable Software Analytics” which
was presented by Hoa, Truyen and Aditya in the year 2018 [1]. This research will include
various sections such as the intention of paper, utilized research methodologies, issues described
in the paper, results, and conclusion.
The intention of the paper
According to the writers, software analytics is a major topic for the investigation and many
companies adopted such technology in the business activities recently that helped to manage
larger datasets effectively. It is determined that most of the software analytics techniques are
based on machine learning programs that help to develop prediction networks for numerous
software engineering tasks including defect prediction, API recommendations, threat
identification and many more. In this paper, the investigators determined the major
characteristics of software analytics and evaluated explainable software analytics models [2].
Using software analytic networks, the companies can enhance the performance of computing
software and the accuracy of predictive models can be enhanced using deep neural networks and
ensemble techniques. It is highlighted that the lack of explainability outcomes in the lack of trust
may lead the business reluctance for adopting software analytics in the workplace. The
researchers argued that the larger predictive performance at the time of testing can develop some
trust in the models which can help to manage larger datasets and perform complex tasks easily.
The major research question of the paper is “how might companies evaluate explainability of
software analytics frameworks and how do they develop explanations from software analytics?”
For addressing such a question, the writers included viewpoints of other investigators and

RESEARCH PAPER 3
provided depth information in regards to the software analytic models [3]. The authors adapted
the model of explanation that combines the software analytics and risk predictive approaches in
the model. It is observed that the major reason for adopting software analytics in the companies
is that software practitioners are reluctant for trusting predictions developed by the analytics
machinery without evaluating the significance of the predictions [4].
This paper provided a way to enhance skills and experiences in the case of explainable software
analytics and examine the significance of software analytics in the risk predictions. Numerous
headings are included in the paper such as good explanation in software engineering, explainable
software analytics, explanations for deep models and many more. The key intention of the
chosen paper is to evaluate the role of software analytic models in the achievement of accurate
predictions.
Research methods
Research approach
Mainly, in the investigations two kinds of approaches could be adopted included inductive and
deductive based on the research concerns. In the chosen paper, an inductive approach is adopted
due to the subjective nature of the topic where it helped to perform in-depth analysis in regards
to the software analytic models. Using such an approach, the writers developed effective plans
and managed the flow of information which helped to address research concerns in an
appropriate manner. So, it is highlighted that the writers included an inductive approach that
provided a way to complete the investigation with better quality and obtain effective data in the
research [5].
Research design
It is a significant section of the investigation that has the capability to provide reliable data in the
investigation by which the research can be done in the right direction. In this article, the
qualitative design is adopted due to its ability to provide effective information and manage the
research gaps easily. Using such design the writers developed and implemented effective plans
and strategies in the investigations by which the developed research questions are addressed
effectively.

RESEARCH PAPER 4
Data collection methods
It is found that data gathering is a helpful methodology that includes numerous techniques and
approaches by which the writers can obtain effective data. In the chosen paper the secondary
methods are adopted that helped to include the theoretical information in the investigation in
regards to the software analytics. A literature review is conducted that evaluated the findings of
recent papers about software analytics and reduced the research gaps effectively. Numerous
sources are included in the data gathering techniques including peer-reviewed papers, online
websites, books and many more. So, using such methods the writers adopted reliable information
about software analytics and achieved the designed aims and goals effectively.
Data analysis
The writers adopted a content analysis that has the ability to provide effective information and
analyze the data easily. The presence of such methods can manage the research gaps and address
the research concerns in an appropriate manner. However, the researchers did not include
statistical analysis techniques due to the subjective nature of the investigations. The uses of
content analysis processes enabled the investigators to conclude effective points and manage the
flow of information effectively. The major problem is that the writers did not include primary
information that produced problems in the research. So, it is reported that the presence of such
methods provided reliable and appropriate techniques to the investigators and completed the
research effectively.
Problems highlighted by the writers
The chosen paper is based on software analytics where the investigators addressed and
highlighted the problems of risk predictions from the computing software. It is reported that
software analytics uses advanced networks and systems that have the potential to control and
manage the risks and issues occurred in the computers effectively [6]. Using explainable
software analytics the companies can solve complex tasks easily and defect the predictions and
risks in an appropriate manner. The researchers believed that developing and implementing
predictive based software analytic models are significant as they help to achieve accurate
predictions and manage the data handling issues easily. After reviewing the paper, it is found that

RESEARCH PAPER 5
the writers addressed the problem of risk prediction by developing explainable software analytic
models.
Results and findings
After reviewing the chosen article, it is found that software analytics are reluctant to trust
predictions developed by the analytics machinery without understanding the rationale for those
predictions. The conducted research shows the effectiveness of software analytics in risk
prediction and performance enhancement [7]. It is determined that software analytic models are
capable to develop effective networks and plans that can help to detect the fraud signals from the
systems and lead the level of risk prediction easily. The writers believed that making predictive
models can help companies to achieve accurate and effective predictions. Explainability must be
a significant measure for evaluating software analytic models or frameworks and can be utilized
for analyzing larger datasets easily [8]. Therefore, it is suggested that companies should develop
and implement explainable software analytics in order to manage and handle the risks and issues
from the computing systems and protect data from risks easily.
Conclusions of the paper
It can be summarized that the implementation of a software analytic model is beneficial for the
companies by which the effectiveness of risk prediction can be enhanced easily and issues can be
addressed in an appropriate manner. The writers found that explainable software analytics can be
used to facilitate individual understanding of machine prediction. In the case of software
analytics AI networks are used that provide a way to perform complex tasks easily and detect the
risks and frauds from the computing devices effectively. The chosen paper is linked to the unit of
assignment where the writers provided their viewpoints about artificial intelligence and
evaluated the significance of explainable software analytics which can help to enhance the
experience and skills of the students.
Conclusion
From this report, it can be concluded that artificial intelligence is the best IT technology by
which software analytic models can be developed that can support companies to address the risks
and issues that occurred in the systems easily. This report evaluated and reviewed a journal

RESEARCH PAPER 6
article based on software analytics and determined that the writers provided depth information
about software analytics. The chosen methods in the paper supported the writers to complete the
investigation with better outcomes and also addressed the research gaps effectively. So, it is
suggested that business communities should implement software analytics for achieving larger
prediction effectiveness.

RESEARCH PAPER 7
References
[1].H.K., Dam, T. Tran and, A., Ghose, “Explainable software analytics,” In Proceedings of
the 40th International Conference on Software Engineering: New Ideas and Emerging
Results, vol. 12, no. 6, pp. 53-56, 2018.
[2].W., Fu, T. Menzies and, X., Shen, “Tuning for software analytics: Is it really
necessary?” Information and Software Technology, vol. 76, no. 6, pp.135-146, 2016.
[3].Y., Yang, D., Falessi, T. Menzies and, J., Hihn, “Actionable analytics for software
engineering,” IEEE Software, vol. 35, no. 1, pp.51-53, 2017.
[4].A., Agrawal, W., Fu, D., Chen, X. Shen and, T., Menzies, “How to" DODGE" Complex
Software Analytics,” IEEE Transactions on Software Engineering, vol. 12, no. 6, pp. 12-
18, 2019.
[5].T. Menzies and, T., Zimmermann, “Software Analytics: What’s Next?,” IEEE
Software, vol. 35, no. 5, pp.64-70, 2018.
[6].T. Menzies and, M., Shepperd, “Bad smells in software analytics papers,” Information
and Software Technology, vol. 112, no. 6, pp.35-47, 2019.
[7].S., Martínez-Fernández, A.M., Vollmer, A., Jedlitschka, X., Franch, L., López, P., Ram,
P., Rodríguez, S., Aaramaa, A., Bagnato, M. Choraś and, J., Partanen, “Continuously
assessing and improving software quality with software analytics tools: a case study,”
IEEE access, vol. 7, no. 4, pp.68219-68239, 2019.
[8].J., Guo, M., Rahimi, J., Cleland-Huang, A., Rasin, J.H. Hayes and, M., Vierhauser,
“Cold-start software analytics,” In Proceedings of the 13th International Conference on
Mining Software Repositories, vol. 12, no. 6, pp. 142-153, 2016.