Search

Technological advances in the past decade, hardware and software alike, have made access to high-performance computing (HPC) easier than ever. We review these advances from a statistical computing perspective. Cloud computing makes access to supercomputers affordable. Deep learning software libraries make programming statistical algorithms easy and enable users to write code once and run it anywhere-from a laptop to a workstation with multiple graphics processing units (GPUs) or a supercomputer in a cloud. Highlighting how these developments benefit statisticians, we review recent optimization algorithms that are useful for high-dimensional models and can harness the power of HPC. Code snippets are provided to demonstrate the ease of programming. We also provide an easy-to-use distributed matrix data structure suitable for HPC. Employing this data structure, we illustrate various statistical applications including large-scale positron emission tomography and ℓ1-regularized Cox regression. Our examples easily scale up to an 8-GPU workstation and a 720-CPU-core cluster in a cloud. As a case in point, we analyze the onset of type-2 diabetes from the UK Biobank with 200,000 subjects and about 500,000 single nucleotide polymorphisms using the HPC ℓ1-regularized Cox regression. Fitting this half-million-variate model takes less than 45 minutes and reconfirms known associations. To our knowledge, this is the first demonstration of the feasibility of penalized regression of survival outcomes at this scale.

As the demand for skilled data scientists has grown, university level statistics and data science courses have become more rigorous in training students to understand and utilize the tools that their future careers will likely require. However, the mechanisms to assess students' use of these tools while they are learning to use them are not well defined. As such, a framework to assess statistical computing actions was created. Using task-based interviews of students who completed a second course in statistics, the framework was used to determine the ways in which students utilize statistical computing tools, specifically R, while going through problem solving phases. Patterns that emerged are discussed. Supplementary materials for this article are available online.

Traditionally, statistical computing courses have taught the syntax of a particular programming language or specific statistical computation methods. Since Nolan and Temple Lang's seminal paper, we have seen a greater emphasis on data wrangling, reproducible research, and visualization. This shift better prepares students for careers working with complex datasets and producing analyses for multiple audiences. But, we argue, statisticians are now often called upon to develop statistical "software," not just analyses, such as R packages implementing new analysis methods or machine learning systems integrated into commercial products. This demands different skills. We describe a graduate course that we developed to meet this need by focusing on four themes: programming practices, software design, important algorithms and data structures, and essential tools and methods. Through code review and revision, and a semester-long software project, students practice all the skills of software engineering. The course allows students to expand their understanding of computing "as applied to statistical problems" while building expertise in the kind of software development that is increasingly the province of the working statistician. We see this as a model for the future evolution of the computing curriculum in statistics and data science.

The field of statistical computing is rapidly developing and evolving. Shifting away from the formerly siloed landscape of mathematics, statistics, and computer science, recent advancements in statistical computing are largely characterized by a fusing of these worlds; namely, programming, software development, and applied statistics are merging in new and exciting ways. There are numerous drivers behind this advancement, including open movement (encompassing development, science, and access), the advent of data science as a field, and collaborative problem-solving, as well as practice-altering advances in subfields such as artificial intelligence, machine learning, and Bayesian estimation. In this paper, we trace this shift in how modern statistical computing is performed, and that which has recently emerged from it. This discussion points to a future of boundless potential for the field.

As the demand for skilled data scientists has grown, university level statistics and data science courses have become more rigorous in training students to understand and utilize the tools that their future careers will likely require. However, the mechanisms to assess students’ use of these tools while they are learning to use them are not well defined. As such, a framework to assess statistical computing actions was created. Using task-based interviews of students who completed a second course in statistics, the framework was used to determine the ways in which students utilize statistical computing tools, specifically R, while going through problem solving phases. Patterns that emerged are discussed. Supplementary materials for this article are available online.

Modern environmental science research increasingly requires computational ability to apply statistics to environmental science problems, but graduate students in these scientific fields typically lack these integral skills. Many scientific graduate degree programs expect students to acquire these computational skills in an applied statistics course. A gap remains, however, between the computational skills required for the implementation of statistics in scientific research and those taught in statistics courses. This qualitative study examines how five environmental science graduate students at one institution experience the phenomenon of acquiring the computational skills necessary to implement statistics in their research and the factors that foster or inhibit learning. In-depth interviews revealed three themes in these students' paths towards computational knowledge acquisition: use of peer support, seeking out a singular "consultant," and learning through independent research experiences. These themes provide rich descriptions of graduate student experiences and strategies used while developing computational skills to apply statistics in their own research, thus informing how to improve instruction, both in and out of the formal classroom.

Traditionally, statistical computing courses have taught the syntax of a particular programming language or specific statistical computation methods. Since Nolan and Temple Lang’s seminal paper, we have seen a greater emphasis on data wrangling, reproducible research, and visualization. This shift better prepares students for careers working with complex datasets and producing analyses for multiple audiences. But, we argue, statisticians are now often called upon to develop statistical software, not just analyses, such as R packages implementing new analysis methods or machine learning systems integrated into commercial products. This demands different skills. We describe a graduate course that we developed to meet this need by focusing on four themes: programming practices, software design, important algorithms and data structures, and essential tools and methods. Through code review and revision, and a semester-long software project, students practice all the skills of software engineering. The course allows students to expand their understanding of computing as applied to statistical problems while building expertise in the kind of software development that is increasingly the province of the working statistician. We see this as a model for the future evolution of the computing curriculum in statistics and data science.

Coronavirus disease 2019 (COVID-19), which causes severe respiratory illness, has become a pandemic. The World Health Organization has declared it a public health crisis of international concern. We developed a susceptible, exposed, infected, recovered (SEIR) model for COVID-19 to show the importance of estimating the reproduction number (R0). This work is focused on predicting the COVID-19 outbreak in its early stage in India based on an estimation of R0. The developed model will help policymakers to take active measures prior to the further spread of COVID-19. Data on daily newly infected cases in India from March 2, 2020 to April 2, 2020 were to estimate R0 using the earlyR package. The maximum-likelihood approach was used to analyze the distribution of R0 values, and the bootstrap strategy was applied for resampling to identify the most likely R0 value. We estimated the median value of R0 to be 1.471 (95% confidence interval [CI], 1.351 to 1.592) and predicted that the new case count may reach 39,382 (95% CI, 34,300 to 47,351) in 30 days.

The field of statistical computing is rapidly developing and evolving. Shifting away from the formerly siloed landscape of mathematics, statistics, and computer science, recent advancements in statistical computing are largely characterized by a fusing of these worlds; namely, programming, software development, and applied statistics are merging in new and exciting ways. There are numerous drivers behind this advancement, including open movement (encompassing development, science, and access), the advent of data science as a field, and collaborative problem-solving, as well as practice-altering advances in subfields such as artificial intelligence, machine learning, and Bayesian estimation. In this paper, we trace this shift in how modern statistical computing is performed, and that which has recently emerged from it. This discussion points to a future of boundless potential for the field. [ABSTRACT FROM AUTHOR]

Biodiversity characters of the landscape provide basis of prioritizing the sites in conservation effort. There is an urgent need for rapid assessment of existing biodiversity using state-of-art tools and technologies at large scale. The purpose of the study is to model and prioritize biological richness based on multi-criteria decision analysis (MCDA) for conservation priority and management planning. Vegetation type map for year 2017 was developed for generation of various landscape indices e.g. fragmentation, patchiness, porosity, juxtaposition etc. The Spatial Biodiversity Model (SBM) prepared for similar landscape of Uttarakhanda, India which is scale, resolution and location independent for spatial biodiversity richness modelling was executed in R programming platform. Satellite data, non-spatial data and ancillary data were used to generate Biological Richness (BR) map which is categorized into 4 classes as low, moderate, high and very high (biodiversity rich) including non-forest area to quantify BR area. The result shows that largest area is under very high biological richness class followed by high, moderate and low BR area. Overall accuracy and Kappa Statistics of LULC/vegetation type classification is 82.61% and 0.8013 respectively. The spatial regression analysis for final output validation has been made with ground based species diversity data where R2 value for Shannon-Wiener index and Margalef’s diversity index are 0.64 and 0.56 respectively. The results also re-emphasize the role of geospatial techniques in the quick appraisal of predicting biological richness. The study result is applicable in systematic inventory of biological resources, land use planning, conservation prioritization and policy support.

In the presence of massive data coming with high heterogeneity we need to change our statistical thinking and statistical education in order to adapt both - classical statistics and software developments that address new challenges. Significant developments include open data, big data, data visualisation, and they are changing the nature of the evidence that is available, the ways in which it is presented and the skills needed for its interpretation. The amount of information is not the most important issue – the real challenge is the combination of the amount and the complexity of data. Moreover, a need arises to know how uncertain situations should be dealt with and what decisions should be taken when information is insufficient (which can also be observed for large datasets). In the paper we discuss the idea of computational statistics as a new approach to statistical teaching and we try to answer a question: how we can best prepare the next generation of statisticians.