Low-Cost Machine Learning for Effective and Efficient Bad Smells Detection

Bad smells are characteristics of software that indicate a code or design problem which can make information system hard to understand, evolve, and maintain. To address this problem, different approaches, manual and automated, have been proposed over the years, including more recently machine learning alternatives. However, despite the advances achieved, some machine learning techniques have not yet been effectively explored, such as the use of feature selection techniques. Moreover, it is not clear to what extent the use of numerous source-code features are necessary for reasonable bad smell detection success. Therefore, in this work we propose an approach using low-cost machine learning for effective and efficient detection of bad smells, through explicit feature selection. Our results showed that the selection allowed to statistically improve the effectiveness of the models. For some cases, the models achieved statistical equivalence, but relying on a highly reduced set of features. Indeed, by using explicit feature selection, simpler models such as Naive Bayes became statistically equivalent to robust models such as Random Forest. Therefore, the selection of features allowed keeping competitive or even superior effectiveness while also improving the efficiency of the models, demanding less computational resources for source-code preprocessing, model training and bad smell detection.

The tiny effects of respiratory masks on physiological, subjective, and behavioral measures under mental load in a randomized controlled trial

Since the outbreak of the coronavirus disease (COVID-19), face coverings are recommended to diminish person-to-person transmission of the SARS-CoV-2 virus. Some public debates concern claims regarding risks caused by wearing face masks, like, e.g., decreased blood oxygen levels and impaired cognitive capabilities. The present, pre-registered study aims to contribute clarity by delivering a direct comparison of wearing an N95 respirator and wearing no face covering. We focused on a demanding situation to show that cognitive efficacy and individual states are equivalent in both conditions. We conducted a randomized-controlled crossover trial with 44 participants. Participants performed the task while wearing an N95 FFR versus wearing none. We measured physiological (blood oxygen saturation and heart rate variability), behavioral (parameters of performance in the task), and subjective (perceived mental load) data to substantiate our assumption as broadly as possible. We analyzed data regarding both statistical equivalence and differences. All of the investigated dimensions showed statistical equivalence given our pre-registered equivalence boundaries. None of the dimensions showed a significant difference between wearing an FFR and not wearing an FFR.Trial Registration: Preregistered with the Open Science Framework: https://osf.io/c2xp5 (15/11/2020). Retrospectively registered with German Clinical Trials Register: DRKS00024806 (18/03/2021).

The 25-Item Ontario Child Health Study Emotional Behavioural Scales-Brief Version (OCHS-EBS-B): Test-Retest Reliability and Construct Validity When Used as Categorical Measures

Objective Child and youth mental health problems are often assessed by parent self-completed checklists that produce dimensional scale scores. When converted to binary ratings of disorder, little is known about their psychometric properties in relation to classifications based on lay-administered structured diagnostic interviews. In addition to estimating agreement, our objective is to test for statistical equivalence in the test-retest reliability and construct validity of two instruments used to classify child emotional, behavioural, and attentional disorders: the 25-item, parent completed Ontario Child Health Study Emotional Behavioural Scales-Brief Version (OCHS-EBS-B) and the Mini International Neuropsychiatric Interview for Children and Adolescents-parent version (MINI-KID-P). Methods This study draws on independent samples ( n = 452) and uses the confidence interval approach to test for statistical equivalence. Reliability is based on kappa (κ). Construct validity is based on standardized beta coefficients (β) estimated in structural equation models. Results The average differences between the MINI-KID-P and OCHS-EBS-B in κ and β were −0.022 and −0.020, respectively. However, in both instances, criteria for statistical equivalence were met in only 5 of 12 comparisons. Based on κ, between-instrument agreement on the classifications of disorder went from 0.481 (attentional disorder) to 0.721 (emotional disorder) but were substantially higher (0.731 to 0.895, respectively) when corrected for attenuation due to measurement error. Conclusions Although falling short of equivalence, the results suggest on balance that the reliability and validity of the two instruments for classifying child psychiatric disorder assessed by parents are highly comparable. This conclusion is supported by the high levels of agreement between the instruments after correcting for attenuation due to measurement error.

Bayesian Hodges-Lehmann tests for statistical equivalence in the two-sample setting: Power analysis, type I error rates and equivalence boundary selection in biomedical research

Background Null hypothesis significance testing (NHST) is among the most frequently employed methods in the biomedical sciences. However, the problems of NHST and p-values have been discussed widely and various Bayesian alternatives have been proposed. Some proposals focus on equivalence testing, which aims at testing an interval hypothesis instead of a precise hypothesis. An interval hypothesis includes a small range of parameter values instead of a single null value and the idea goes back to Hodges and Lehmann. As researchers can always expect to observe some (although often negligibly small) effect size, interval hypotheses are more realistic for biomedical research. However, the selection of an equivalence region (the interval boundaries) often seems arbitrary and several Bayesian approaches to equivalence testing coexist. Methods A new proposal is made how to determine the equivalence region for Bayesian equivalence tests based on objective criteria like type I error rate and power. Existing approaches to Bayesian equivalence testing in the two-sample setting are discussed with a focus on the Bayes factor and the region of practical equivalence (ROPE). A simulation study derives the necessary results to make use of the new method in the two-sample setting, which is among the most frequently carried out procedures in biomedical research. Results Bayesian Hodges-Lehmann tests for statistical equivalence differ in their sensitivity to the prior modeling, power, and the associated type I error rates. The relationship between type I error rates, power and sample sizes for existing Bayesian equivalence tests is identified in the two-sample setting. Results allow to determine the equivalence region based on the new method by incorporating such objective criteria. Importantly, results show that not only can prior selection influence the type I error rate and power, but the relationship is even reverse for the Bayes factor and ROPE based equivalence tests. Conclusion Based on the results, researchers can select between the existing Bayesian Hodges-Lehmann tests for statistical equivalence and determine the equivalence region based on objective criteria, thus improving the reproducibility of biomedical research.

Sub-horizon modes and growth index in a linear scalar cosmological perturbations

We test a four dimensional cosmological model embedded in a five dimensional bulk space by means of the dynamical Nash-Greene theorem. In a fluid approach, we apply a joint likelihood analysis to the data with the Markov Chain Monte Carlo (MCMC) method for cosmological parameter estimation. We use recent datasets as the “Gold 2018” growth-rate data, the Planck2018/$$\varLambda $$ Λ CDM data on the cosmic microwave background (CMB) anisotropies, the Baryon acoustic oscillations (BAO) measurements, the Pantheon Supernovae type Ia and the data on the Hubble parameter H(z) with redshift ranging from $$0.01< z < 2.3$$ 0.01 < z < 2.3 . Performing the Information Criterion (IC) analysis, we find that the present model is in very good agreement with observations with a close statistical equivalence with wCDM cosmologies at 1-$$\sigma $$ σ level with a slightly larger growth profiles. By modifications of () code, we make a comparison between the models on their unlensed CMB TT temperature spectra. Moreover, the proposed model presents a low power spectrum by the reduction of the ISW effect at lower multipoles. We also find that the overall percentage relative difference of the growth index $$\varDelta \gamma (\%)$$ Δ γ ( % ) is up to 1.4$$\%$$ % as compared to wCDM pattern in sub-horizon scales.

Wijsman Asymptotic Lacunary I_2-Invariant Equivalence for Double Set Sequences

In this study, for double set sequences, we present the notions of Wijsman asymptotic lacunary invariant equivalence, Wijsman asymptotic lacunary I_2-invariant equivalence and Wijsman asymptotic lacunary I_2^*-invariant equivalence. Also, we examine the relations between these notions and Wijsman asymptotic lacunary invariant statistical equivalence studied in this field before.

Extending greedy feature selection algorithms to multiple solutions

Most feature selection methods identify only a single solution. This is acceptable for predictive purposes, but is not sufficient for knowledge discovery if multiple solutions exist. We propose a strategy to extend a class of greedy methods to efficiently identify multiple solutions, and show under which conditions it identifies all solutions. We also introduce a taxonomy of features that takes the existence of multiple solutions into account. Furthermore, we explore different definitions of statistical equivalence of solutions, as well as methods for testing equivalence. A novel algorithm for compactly representing and visualizing multiple solutions is also introduced. In experiments we show that (a) the proposed algorithm is significantly more computationally efficient than the TIE* algorithm, the only alternative approach with similar theoretical guarantees, while identifying similar solutions to it, and (b) that the identified solutions have similar predictive performance.

Plasma androgens and the presence and course of depression in a large cohort of women

Major depressive disorder (MDD) has a higher prevalence in women with supraphysiologic androgen levels. Whether there is also an association between depression and androgen levels in the physiological range, is unknown. This study examined if women with current MDD have higher androgen levels compared to women who have never had MDD, and if androgen levels are associated with onset and remission of MDD. In 1659 women (513 current MDD, 754 remitted MDD, and 392 never MDD), baseline plasma levels of total testosterone, 5α-dihydrotestosterone, and androstenedione were determined with liquid chromatography-tandem mass spectrometry, and dehydroepiandrosterone-sulfate and sex hormone binding globulin (SHBG) with radioimmunoassays. Free testosterone was calculated. MDD status was assessed at baseline, and at 2 and 4 years follow-up. Women were aged between 18 and 65 years (mean age 41) with total testosterone levels in the physiological range (geometric mean 0.72 nmol/L [95% CI 0.27–1.93]). After adjusting for covariates and multiple testing, women with current MDD had a higher mean free testosterone than women who never had MDD (adjusted geometric mean 8.50 vs. 7.55 pmol/L, p = 0.0005), but this difference was not large enough to be considered clinically meaningful as it was consistent with statistical equivalence. Levels of other androgens and SHBG did not differ and were also statistically equivalent between the groups. None of the androgens or SHBG levels predicted onset or remission of MDD. Our findings support the idea that plasma androgens within the physiological range have no or only limited effects on depressive disorders in women.