Reproduction Package for the Paper "How do Machine Learning Projects use Continuous Integration Practices? An Empirical Study on GitHub Actions"This reproduction package contains the necessary materials to replicate the findings presented in the paper published at the Mining Software Repositories (MSR) conference in 2024.Folder Structure:- datasets: Contains all datasets used in the analysis of the Research Questions (RQs) of the paper.- plots: Contains plots generated to present the results of the investigated RQs of the study.- r-script: Contains all R scripts used in the analysis of the RQs, as well as scripts used to compute metrics such as build duration, time to fix broken builds, and test coverage.- RQ3-neovis-network-graph: Contains a README.txt file providing instructions to create the network graph used to present the results of RQ3 using the neovisjs library.Please refer to the specific folders for detailed information on how to reproduce the analysis and results presented in the paper.Furthermore, the code we used to retrieve data for the studied projects is available in the following GitHub repository: https://github.com/joaohelis/ml-ci-project-miner

Reproduction Package for the Paper "How do Machine Learning Projects use Continuous Integration Practices? An Empirical Study on GitHub Actions"This reproduction package contains the necessary materials to replicate the findings presented in the paper published at the Mining Software Repositories (MSR) conference in 2024.Folder Structure:- datasets: Contains all datasets used in the analysis of the Research Questions (RQs) of the paper.- plots: Contains plots generated to present the results of the investigated RQs of the study.- r-script: Contains all R scripts used in the analysis of the RQs, as well as scripts used to compute metrics such as build duration, time to fix broken builds, and test coverage.- RQ3-neovis-network-graph: Contains a README.txt file providing instructions to create the network graph used to present the results of RQ3 using the neovisjs library.Please refer to the specific folders for detailed information on how to reproduce the analysis and results presented in the paper.Furthermore, the code we used to retrieve data for the studied projects is available in the following GitHub repository: https://github.com/joaohelis/ml-ci-project-miner

Aim: To evaluate how the area of ​habitat island systems influences multiple facets of diversity.  Location: Southern Brazil.  Taxon: Birds.  Methods: Using an Information Theoretic approach, we compared the fit of 20 diversity–area relationship (DARs) models in three habitat island systems. We tested for the best-fit model, model-family, shape, and presence/absence of an asymptote. We used species richness (SR), Faith's phylogenetic diversity (PD) and Faith's functional diversity (FD) to assess species–area (SARs), phylogenetic (PDARs) and functional diversity–area relationships (FDARs). We controlled for the effect of SR in PD and FD via null models to assess PDARs and FDARs independently of SR and to explore the influence of phylogenetic and functional randomness, clustering and overdispersion.  Results: PDARs and FDARs built with PD and FD resembled SARs and were all best fitted by convex or sigmoidal, upwards oriented, non-asymptotic models. Controlling for SR in diversity indices produced flat or downwards-oriented, weak PDARs and FDARs, which were best fitted by convex, non-asymptotic models or the linear model. Taxonomic diversity accumulated faster with area than functional diversity, which accumulated faster than phylogenetic diversity. Randomness and clustering patterns prevailed in shaping PDARs and FDARs relative to overdispersion.  Main conclusions: Controlling for SR in PD and FD affects DARs patterns and the strength of the relationships. Irrespective of this influence of SR, a few simple models of the power and exponential model-families best fit DARs. Model parameters reveal differences in the response of each facet of diversity to increases in area, highlighting the complementary nature of DARs. When considered independently of SR, both PDAR and FDAR patterns largely reflect the broad variation of phylogenetic and functional diversity in small islands. It is therefore likely that the ecological processes that promote phylogenetic and functional overdispersion and clustering operate at contrasting spatial scales.