Jelena Vlasenko

Failure prediction based on log files using Random Indexing and Support Vector Machines

Research problem: The impact of failures on software systems can be substantial since the recovery process can require unexpected amounts of time and resources. Accurate failure predictions can help in mitigating the impact of failures. Resources, applications, and services can be scheduled to limit the impact of failures. However, providing accurate predictions sufficiently ahead is challenging. Log files contain messages that represent a change of system state. A sequence or a pattern of messages may be used to predict failures. Contribution: We describe an approach to predict failures based on log files using Random Indexing (RI) and Support Vector Machines (SVMs). Method: RI is applied to represent sequences: each operation is characterized in terms of its context. SVMs associate sequences to a class of failures or non-failures. Weighted SVMs are applied to deal with imbalanced datasets and to improve the true positive rate. We apply our approach to log files collected during approximately three months of work in a large European manufacturing company. Results: According to our results, weighted SVMs sacrifice some specificity to improve sensitivity. Specificity remains higher than 0.80 in four out of six analyzed applications. Conclusions: Overall, our approach is very reliable in predicting both failures and non-failures.

Pair Programming and Software Defects--A Large, Industrial Case Study

In the last decade, there has been increasing interest in pair programming (PP). However, despite the existing work, there is still a lack of substantial evidence of the effects of PP in industrial environments. To address this issue, we have analyzed the work of a team of 17 industrial developers for 14 months. The team is part of the IT department of a large Italian manufacturing company; it adopts a customized version of extreme programming (XP). We have investigated the effects of PP on software quality in five different scenarios. The results show that PP appears to provide a perceivable but small effect on the reduction of defects in these settings.

Predicting Development Effort from User Stories

In this paper, we propose a method for predicting development effort based on user stories. Such approach is well suited for Agile software projects where requirements are developed along with the project and only sketched in a rough manner. We apply the proposed method to two industrial Agile software projects of very different size and structure. We show that such effort estimation works reasonably well if user stories are written in a structured way.

Analysing the Usage of Tools in Pair Programming Sessions

In this study we observe the daily work of nineteen software developers of a large Italian manufacturing company for a period of ten months to determine the effects of pair programming on the use of tools. Fifteen developers are existing team members and four have recently joined the team. They practice Pair Programming spontaneously, that is, when they feel it is needed. We identify the tools the developers use, how they distribute their time among these tools when working alone and when doing Pair Programming. The data have been extracted non-invasively by means of PROM – tool for automated data collection and analysis. The preliminary results indicate that developers working in pairs devote significantly more time to programming activities than developers working alone.

Open Source Software: Mobile Open Source Technologies

During the last years, most of the large free / open source software projects have included code review as an usual, or even mandatory practice for changes to their code. In many cases it is implemented as a process in which a developer proposing some change needs to ask for a review by another developer before it can enter the code base. Code reviews, therefore, become a critical process for the project, which could cause delays in contributions being accepted, and risk to become a bottleneck if not enough reviewers are available. In this paper we present a methodology designed to analyze the code review process, to determine its main characteristics and parameters, and to detect potential problems with it. We also present how we have applied this methodology to the WebKit project, learning about the main characteristics of how code review works in their case.

A Novel Application of Open Source Technologies to Measure Agile Software Development Process

In the last 10 years Open Source products have been widely used in industry. New methodologies and best practices to develop Open Source software appeared. In this work, we present an application that runs on Android-based mobile phones and collects proximity data with other devices via Bluetooth. The application gives new insights into measuring proximity inside a team of software developers. Data collection process is automatic so that the team members are not distracted from their daily activities. The collected data represent time frames when developers work alone at their machines and when they do Pair Programming with their colleagues.

Analyzing tool usage to understand to what extent experts change their activities when mentoring

Automated In-Process Software Engineering Measurement and Analysis (AISEMA) systems represent a major advancement in tracking non-invasively the activities of developers. We have built on the top of an AISEMA system a model that enables to understand better how the tools are used in practical real-life development settings. In this work we evaluate to what extent experienced developers change their activities during mentoring activities in Pair Programming (PP) and, in this case, how long this effect can be observed. We compare how the experienced developers use the tools when working with other experts and when with new developers. The results indicate that there is a notable difference in the way the tools are used between the experts working together and the experts mentoring the new developers that have just joined the team. Moreover, over time the difference between pairs of experts and mixed pairs (experts and novices) working together becomes almost unnoticeable.