Simone Romano

Maria Teresa Baldassarre, Danilo Caivano, Davide Fucci et al.

In this paper, we investigate the effect of TDD, as compared to a non-TDD approach, as well as its retainment (or retention) over a time span of (about) six months. To pursue these objectives, we conducted a (quantitative) longitudinal cohort study with 30 novice developers (i.e., third-year undergraduate students in Computer Science). We observed that TDD affects neither the external quality of software products nor developers' productivity. However, we observed that the participants applying TDD produced significantly more tests, with a higher fault-detection capability than those using a non-TDD approach. As for the retainment of TDD, we found that TDD is retained by novice developers for at least six months.

Adrian Santos, Sira Vegas, Oscar Dieste et al.

Context: Test-driven development (TDD) is an agile software development approach that has been widely claimed to improve software quality. However, the extent to which TDD improves quality appears to be largely dependent upon the characteristics of the study in which it is evaluated (e.g., the research method, participant type, programming environment, etc.). The particularities of each study make the aggregation of results untenable. Objectives: The goal of this paper is to: increase the accuracy and generalizability of the results achieved in isolated experiments on TDD, provide joint conclusions on the performance of TDD across different industrial and academic settings, and assess the extent to which the characteristics of the experiments affect the quality-related performance of TDD. Method: We conduct a family of 12 experiments on TDD in academia and industry. We aggregate their results by means of meta-analysis. We perform exploratory analyses to identify variables impacting the quality-related performance of TDD. Results: TDD novices achieve a slightly higher code quality with iterative test-last development (i.e., ITL, the reverse approach of TDD) than with TDD. The task being developed largely determines quality. The programming environment, the order in which TDD and ITL are applied, or the learning effects from one development approach to another do not appear to affect quality. The quality-related performance of professionals using TDD drops more than for students. We hypothesize that this may be due to their being more resistant to change and potentially less motivated than students. Conclusion: Previous studies seem to provide conflicting results on TDD performance (i.e., positive vs. negative, respectively). We hypothesize that these conflicting results may be due to different study durations, experiment participants being unfamiliar with the TDD process...

Cristiano Politowski, Foutse Khomh, Simone Romano et al.

Context: Several studies investigated the impact of anti-patterns (i.e., "poor" solutions to recurring design problems) during maintenance activities and reported that anti-patterns significantly affect the developers' effort required to edit files. However, before developers edit files, they must understand the source code of the systems. This source code must be easy to understand by developers. Objective: In this work, we provide a complete assessment of the impact of two instances of two anti-patterns, Blob or Spaghetti Code, on program comprehension. Method: We analyze the impact of these two anti-patterns through three empirical studies conducted at Polytechnique Montréal (Canada) with 24 participants; at Carlton University (Canada) with 30 participants; and at University Basilicata (Italy) with 79 participants. Results: We collect data from 372 tasks obtained thanks to 133 different participants from the three universities. We use three metrics to assess the developers' comprehension of the source code: (1) the duration to complete each task; (2) their percentage of correct answers; and, (3) the NASA task load index for their effort. Conclusions: We report that, although single occurrences of Blob or Spaghetti code anti-patterns have little effect on code comprehension, two occurrences of either Blob or Spaghetti Code significantly increases the developers' time spent in their tasks, reduce their percentage of correct answers, and increase their effort. Hence, we recommend that developers act on both anti-patterns, which should be refactored out of the source code whenever possible. We also recommend further studies on combinations of anti-patterns rather than on single anti-patterns one at a time.

Simone Romano, Davide Fucci, Giuseppe Scanniello et al.

Researcher Bias (RB) occurs when researchers influence the results of an empirical study based on their expectations.RB might be due to the use of Questionable Research Practices(QRPs). In research fields like medicine, blinding techniques have been applied to counteract RB. We conducted an explorative qualitative survey to investigate RB in Software Engineering (SE)experiments, with respect to (i) QRPs potentially leading to RB, (ii) causes behind RB, and (iii) possible actions to counteract including blinding techniques. Data collection was based on semi-structured interviews. We interviewed nine active experts in the empirical SE community. We then analyzed the transcripts of these interviews through thematic analysis. We found that some QRPs are acceptable in certain cases. Also, it appears that the presence of RB is perceived in SE and, to counteract RB, a number of solutions have been highlighted: some are intended for SE researchers and others for the boards of SE research outlets.

Xingjun Ma, Hanxun Huang, Yisen Wang et al.

Robust loss functions are essential for training accurate deep neural networks (DNNs) in the presence of noisy (incorrect) labels. It has been shown that the commonly used Cross Entropy (CE) loss is not robust to noisy labels. Whilst new loss functions have been designed, they are only partially robust. In this paper, we theoretically show by applying a simple normalization that: any loss can be made robust to noisy labels. However, in practice, simply being robust is not sufficient for a loss function to train accurate DNNs. By investigating several robust loss functions, we find that they suffer from a problem of underfitting. To address this, we propose a framework to build robust loss functions called Active Passive Loss (APL). APL combines two robust loss functions that mutually boost each other. Experiments on benchmark datasets demonstrate that the family of new loss functions created by our APL framework can consistently outperform state-of-the-art methods by large margins, especially under large noise rates such as 60% or 80% incorrect labels.

Simone Romano, Giuseppe Scanniello, Maria Teresa Baldassarre et al.

Test-driven Development (TDD) is an incremental approach to software development. Despite it is claimed to improve both quality of software and developers' productivity, the research on the claimed effects of TDD has so far shown inconclusive results. Some researchers have ascribed these inconclusive results to the negative affective states that TDD would provoke. A previous (baseline) experiment has, therefore, studied the affective reactions of (novice) developers---i.e., 29 third-year undergraduates in Computer Science (CS)---when practicing TDD to implement software. To validate the results of the baseline experiment, we conducted a replicated experiment that studies the affective reactions of novice developers when applying TDD to develop software. Developers in the treatment group carried out a development task using TDD, while those in the control group used a non-TDD approach. To measure the affective reactions of developers, we used the Self-Assessment Manikin instrument complemented with a liking dimension. The most important differences between the baseline and replicated experiments are: (i) the kind of novice developers involved in the experiments---third-year vs. second-year undergraduates in CS from two different universities; and (ii) their number---29 vs. 59. The results of the replicated experiment do not show any difference in the affective reactions of novice developers. Instead, the results of the baseline experiment suggest that developers seem to like TDD less as compared to a non-TDD approach and that developers following TDD seem to like implementing code less than the other developers, while testing code seems to make them less happy.

Simone Romano, Davide Fucci, Maria Teresa Baldassarre et al.

We study whether and in which phase Test-Driven Development (TDD) influences affective states of novice developers in terms of pleasure, arousal, dominance, and liking. We performed a controlled experiment with 29 novice developers. Developers in the treatment group performed a development task using TDD, whereas those in the control group used a non-TDD development approach. We compared the affective reactions to the development approaches, as well as to the implementation and testing phases, exploiting a lightweight, powerful, and widely used tool, i.e., Self-Assessment Manikin. We observed that there is a difference between the two development approaches in terms of affective reactions. Therefore, it seems that affective reactions play an important role when applying TDD and their investigation could help researchers to better understand such a development approach

Simone Romano, Giuseppe Scanniello, Davide Fucci et al.

Background: Noise, defined as an unwanted sound, is one of the commonest factors that could affect people's performance in their daily work activities. The software engineering research community has marginally investigated the effects of noise on software engineers' performance. Aims: We studied if noise affects software engineers' performance in (i) comprehending functional requirements and (ii) fixing faults in the source code. Method: We conducted two experiments with final-year undergraduate students in Computer Science. In the first experiment, we asked 55 students to comprehend functional requirements exposing them or not to noise, while in the second experiment 42 students were asked to fix faults in Java code. Results: The participants in the second experiment, when exposed to noise, had significantly worse performance in fixing faults in the source code. On the other hand, we did not observe any statistically significant difference in the first experiment. Conclusions: Fixing faults in source code seems to be more vulnerable to noise than comprehending functional requirements.

Davide Fucci, Simone Romano, Maria Teresa Baldassarre et al.

Background: Test-Driven Development (TDD) is an agile software development practice, which is claimed to boost both external quality of software products and developers' productivity. Aims: We want to study (i) the TDD effects on the external quality of software products as well as the developers' productivity, and (ii) the retainment of TDD over a period of five months. Method: We conducted a (quantitative) longitudinal cohort study with 30 third year undergraduate students in Computer Science at the University of Bari in Italy. Results: The use of TDD has a statistically significant effect neither on the external quality of software products nor on the developers' productivity. However, we observed that participants using TDD produced significantly more tests than those applying a non-TDD development process and that the retainment of TDD is particularly noticeable in the amount of tests written. Conclusions: Our results should encourage software companies to adopt TDD because who practices TDD tends to write more tests---having more tests can come in handy when testing software systems or localizing faults---and it seems that novice developers retain TDD.

Davide Fucci, Giuseppe Scanniello, Simone Romano et al.

We present a quasi-experiment to investigate whether, and to what extent, sleep deprivation impacts the performance of novice software developers using the agile practice of test-first development (TFD). We recruited 45 undergraduates and asked them to tackle a programming task. Among the participants, 23 agreed to stay awake the night before carrying out the task, while 22 slept usually. We analyzed the quality (i.e., the functional correctness) of the implementations delivered by the participants in both groups, their engagement in writing source code (i.e., the amount of activities performed in the IDE while tackling the programming task) and ability to apply TFD (i.e., the extent to which a participant can use this practice). By comparing the two groups of participants, we found that a single night of sleep deprivation leads to a reduction of 50% in the quality of the implementations. There is important evidence that the developers' engagement and their prowess to apply TFD are negatively impacted. Our results also show that sleep-deprived developers make more fixes to syntactic mistakes in the source code. We conclude that sleep deprivation has possibly disruptive effects on software development activities. The results open opportunities for improving developers' performance by integrating the study of sleep with other psycho-physiological factors in which the software engineering research community has recently taken an interest in.

Yang Lei, James C. Bezdek, Simone Romano et al.

It has been noticed that some external CVIs exhibit a preferential bias towards a larger or smaller number of clusters which is monotonic (directly or inversely) in the number of clusters in candidate partitions. This type of bias is caused by the functional form of the CVI model. For example, the popular Rand index (RI) exhibits a monotone increasing (NCinc) bias, while the Jaccard Index (JI) index suffers from a monotone decreasing (NCdec) bias. This type of bias has been previously recognized in the literature. In this work, we identify a new type of bias arising from the distribution of the ground truth (reference) partition against which candidate partitions are compared. We call this new type of bias ground truth (GT) bias. This type of bias occurs if a change in the reference partition causes a change in the bias status (e.g., NCinc, NCdec) of a CVI. For example, NCinc bias in the RI can be changed to NCdec bias by skewing the distribution of clusters in the ground truth partition. It is important for users to be aware of this new type of biased behaviour, since it may affect the interpretations of CVI results. The objective of this article is to study the empirical and theoretical implications of GT bias. To the best of our knowledge, this is the first extensive study of such a property for external cluster validity indices.

Simone Romano, Nguyen Xuan Vinh, James Bailey et al.

Adjusted for chance measures are widely used to compare partitions/clusterings of the same data set. In particular, the Adjusted Rand Index (ARI) based on pair-counting, and the Adjusted Mutual Information (AMI) based on Shannon information theory are very popular in the clustering community. Nonetheless it is an open problem as to what are the best application scenarios for each measure and guidelines in the literature for their usage are sparse, with the result that users often resort to using both. Generalized Information Theoretic (IT) measures based on the Tsallis entropy have been shown to link pair-counting and Shannon IT measures. In this paper, we aim to bridge the gap between adjustment of measures based on pair-counting and measures based on information theory. We solve the key technical challenge of analytically computing the expected value and variance of generalized IT measures. This allows us to propose adjustments of generalized IT measures, which reduce to well known adjusted clustering comparison measures as special cases. Using the theory of generalized IT measures, we are able to propose the following guidelines for using ARI and AMI as external validation indices: ARI should be used when the reference clustering has large equal sized clusters; AMI should be used when the reference clustering is unbalanced and there exist small clusters.

Simone Romano, Nguyen Xuan Vinh, James Bailey et al.

Estimating the strength of dependency between two variables is fundamental for exploratory analysis and many other applications in data mining. For example: non-linear dependencies between two continuous variables can be explored with the Maximal Information Coefficient (MIC); and categorical variables that are dependent to the target class are selected using Gini gain in random forests. Nonetheless, because dependency measures are estimated on finite samples, the interpretability of their quantification and the accuracy when ranking dependencies become challenging. Dependency estimates are not equal to 0 when variables are independent, cannot be compared if computed on different sample size, and they are inflated by chance on variables with more categories. In this paper, we propose a framework to adjust dependency measure estimates on finite samples. Our adjustments, which are simple and applicable to any dependency measure, are helpful in improving interpretability when quantifying dependency and in improving accuracy on the task of ranking dependencies. In particular, we demonstrate that our approach enhances the interpretability of MIC when used as a proxy for the amount of noise between variables, and to gain accuracy when ranking variables during the splitting procedure in random forests.