Focusing on the Needs: Experiences of Developing a Data Science Program

Donoho’s article “50 Years of Data Science” is a well-thought explanation of a newly developed discipline called “data science.” In this article, we examine his explanations and suggestions about data science, follow-up on some of the issues he mentioned, and share our experiences in developing a data science curriculum and the teaching of related courses.     

Information Geometry and Interior-Point Algorithms in Semidefinite Programs and Symmetric Cone Programs

We develop an information geometric approach to conic programming. Information geometry is a differential geometric framework specifically tailored to deal with convexity, naturally arising in information science including statistics, machine learning and signal processing etc. First we introduce an information geometric framework of conic programming. Then we focus on semidefinite and symmetric cone programs. Recently, we demonstrated that the number of iterations of Mizuno–Todd–Ye predictor–corrector primal–dual interior-point methods is (asymptotically) expressed with an integral over the central trajectory called “the curvature integral”. The number of iterations of the algorithm is approximated surprisingly well with the integral even for fairly large linear/semidefinite programs with thousands of variables. Here we prove that “the curvature integral” admits a rigorous differential geometric expression based on information geometry. We also obtain an interesting information geometric global theorem on the central trajectory for linear programs. Together with the numerical evidence in the aforementioned work, we claim that “the number of iterations of the interior-point algorithm is expressed as a differential geometric quantity.”     

The Effects of Project‐Based Learning on Students' Academic Achievement,Attitude, and Retention of Knowledge: The Subject of “Electricity in Our Lives”

The aim of this study is to analyze the effects of project‐based learning on students' academic achievement, attitude, and retention of knowledge in relation to the subject of “Electricity in Our Lives” in a fourth‐grade science course. The study was conducted in a quasi‐experimental design as a “pre‐test, post‐test with control group.” In the experimental group, the unit was taught through the project‐based learning method. The measuring tools were administered to both groups before and after the applications. To perfectly analyze the “process” of the method, seven different learning assessment “forms” were administered to the students. The findings of the forms indicated that the students learn to construct their own learning and to evaluate changes in their own behavior through the application of the method. The application of different methods between both groups had a statistically significant effect in terms of academic achievement, (F(1,112) = 46.78, p = .000) and of retention of knowledge (F(1,112) = 35.24, p = .000). However, there were no statistically significant effects from being in different groups for the attitudes of students (F(1,112) = .99, p = .321). For the students, being in the project‐based learning groups resulted in better academic achievement and retention of knowledge than being in the traditional teaching group.     

Students' Attitudes Toward Science as Predictors of Gains on Student Content Knowledge: Benefits of an After‐School Program

High‐quality after‐school programs devoted to science have the potential to enhance students' science knowledge and attitudes, which may impact their decisions about pursuing science‐related careers. Because of the unique nature of these informal learning environments, an understanding of the relationships among aspects of students' content knowledge acquisition and attitudes toward science may aid in the development of effective science‐related interventions. We investigated the impact of a semester‐long after‐school intervention utilizing an inquiry‐based infectious diseases curriculum (designed for use after‐school) on 63 urban students' content knowledge and aspects of their attitudes toward science. Content knowledge increased 24.6% from pretest to posttest. Multiple regression analyses indicated suggested that the “self‐directed effort” subscale of the Simpson–Troost Attitude Questionnaire—Revised best predicted increases in students' science content knowledge. The construct “science is fun for me” served as a suppressor effect. These findings suggest that future after‐school programs focusing on aspects of attitudes toward science most closely associated with gains in content knowledge might improve students' enthusiasm and academic preparedness for additional science coursework by improving student attitudes toward their perceptions of their self‐directed effort.     

Statistician Heal Thyself: Have We Lost the Plot?

In 1984, Cleveland suggested that statisticians have an important role in changing the use of graphics in science for the better. Thirty years later, we compared graphs published in top-rated applied science and statistics journals, evaluated for overall quality and against five principles of graphical excellence. Nearly 40% of the 97 graphs we sampled were rated as poor, with no striking differences between the applied science and statistics graphs. Better use of graphs requires better definition of variables, units of measurement, scales, groups, and other graphical elements, and more routine use of grid lines on a “standard” set of graphical forms. Progress over the next 30 years needs to be supported by changes in software defaults.     

Nonparametric universal copula modeling

To handle the ubiquitous problem of “dependence learning,” copulas are quickly becoming a pervasive tool across a wide range of data‐driven disciplines encompassing neuroscience, finance, econometrics, genomics, social science, machine learning, healthcare, and many more. At the same time, despite their practical value, the empirical methods of “learning copula from data” have been unsystematic with full of case‐specific recipes. Taking inspiration from modern LP‐nonparametrics, this paper presents a modest contribution to the need for a more unified and structured approach of copula modeling that is simultaneously valid for arbitrary combinations of continuous and discrete variables.     

Using Prompted Praxis to Improve Teacher Professional Development in Culturally Diverse Schools

Recent science and teacher education reports continue to stress the need for radical changes in the way teachers are prepared to teach science to diverse learners. In response, a three‐year intervention project was developed to help teachers in culturally diverse schools transform their science teaching practices using learning technologies. Many challenges arose that called for strategies to further manage the progress of the project. This paper describes how one of those strategies, “prompted praxis,” was used to manage two of the main challenges encountered, the teachers' following through with their professional development goals and our own sense of urgency to effect change.     

“统计信息”与“信息高速公路”

本文从我国统计信息事业的现状，提出了关于第一类统计与第二类统计。同时指出统计信息事业的迅速发展与繁荣，首先必须实现两类统计的相互渗透；其二必须与计算机技术密切结合，以建立起各级各类统计信息的高速公路，使统计信息更好地为我国的科技、经济、社会的全面发展提供有效的服务。     

Interior hölder estimates for solutions of schrödinger equations and the regularity of nodal sets

Abstract

Motivated by the study of selfdual vortices in gauge field theory, we consider a class of Mean Field equations of Liouville-type on compact surfaces involving singular data assigned by Dirac measures supported at finitely many points (the so called vortex points). According to the applications, we need to describe the blow-up behavior of solution-sequences which concentrate exactly at the given vortex points. We provide accurate pointwise estimates for the profile of the bubbling sequences as well as “sup + inf” estimates for solutions. Those results extend previous work of Li [Li, Y. Y. (1999). Harnack type inequality: The method of moving planes. Comm. Math. Phys. 200:421–444] and Brezis et al. [Brezis, H., Li, Y. Shafrir, I. (1993). A sup + inf inequality for some nonlinear elliptic equations involving the exponential nonlinearities. J. Funct. Anal. 115: 344–358] relative to the “regular” case, namely in absence of singular sources.     

Beginning Secondary Science Teachers' Conceptualization and Enactment of Inquiry‐Based Instruction

This study investigates the conceptions and use of inquiry during classroom instruction among beginning secondary science teachers. The 44 participants were beginning secondary science teachers in their first year of teaching. In order to capture the participants' conceptions of inquiry, the teachers were interviewed and observed during the school year. The interviews consisted of questions about inquiry instruction, while the observations documented the teachers' use of inquiry. All of the interviews were transcribed or coded in order to understand the conceptions of inquiry held by the teachers, and all of the observations were analyzed in order to determine the presence of inquiry during the lesson. The standard for assessing inquiry came from the National Science Education Standards. A quantitative analysis of the data indicated that the teachers frequently talked about implementing “scientific questions” and giving “priority to evidence.” This study found a consistency between the way new teachers talked about inquiry and the way they practiced it in their classrooms. Overall, our observations and interviews revealed that the beginning secondary science teachers tended to enact teacher‐centered forms of inquiry, and could benefit from induction programs focused on inquiry instruction.     

Learning from imprecise and fuzzy observations: Data disambiguation through generalized loss minimization

Methods for analyzing or learning from “fuzzy data” have attracted increasing attention in recent years. In many cases, however, existing methods (for precise, non-fuzzy data) are extended to the fuzzy case in an ad-hoc manner, and without carefully considering the interpretation of a fuzzy set when being used for modeling data. Distinguishing between an ontic and an epistemic interpretation of fuzzy set-valued data, and focusing on the latter, we argue that a “fuzzification” of learning algorithms based on an application of the generic extension principle is not appropriate. In fact, the extension principle fails to properly exploit the inductive bias underlying statistical and machine learning methods, although this bias, at least in principle, offers a means for “disambiguating” the fuzzy data. Alternatively, we therefore propose a method which is based on the generalization of loss functions in empirical risk minimization, and which performs model identification and data disambiguation simultaneously. Elaborating on the fuzzification of specific types of losses, we establish connections to well-known loss functions in regression and classification. We compare our approach with related methods and illustrate its use in logistic regression for binary classification.     

A General Framework for Mining Massive Data Streams

In many domains, data now arrive faster than we are able to mine it. To avoid wasting these data, we must switch from the traditional “one-shot” data mining approach to systems that are able to mine continuous, high-volume, open-ended data streams as they arrive. In this article we identify some desiderata for such systems, and outline our framework for realizing them. A key property of our approach is that it minimizes the time required to build a model on a stream while guaranteeing (as long as the data are iid) that the model learned is effectively indistinguishable from the one that would be obtained using infinite data. Using this framework, we have successfully adapted several learning algorithms to massive data streams, including decision tree induction, Bayesian network learning, k-means clustering, and the EM algorithm for mixtures of Gaussians. These algorithms are able to process on the order of billions of examples per day using off-the-shelf hardware. Building on this, we are currently developing software primitives for scaling arbitrary learning algorithms to massive data streams with minimal effort.     

What are the actual costs of cyber risk events?

Cyber risks are high on the business agenda of every company, but they are difficult to assess due to the absence of reliable data and thorough analyses. This paper is the first to consider a broad range of cyber risk events and actual cost data. For this purpose, we identify cyber losses from an operational risk database and analyze these with methods from statistics and actuarial science. We use the peaks-over-threshold method from extreme value theory to identify “cyber risks of daily life” and “extreme cyber risks”. Human behavior is the main source of cyber risk and cyber risks are very different compared with other risk categories. Our models can be used to yield consistent risk estimates, depending on country, industry, size, and other variables. The findings of the paper are also useful for practitioners, policymakers and regulators in improving the understanding of this new type of risk.     

Graph homomorphisms,the Tutte polynomial and “q-state Potts uniqueness”

We establish for which weighted graphs H homomorphism functions from multigraphs G to H are specializations of the Tutte polynomial of G, answering a question of Freedman, Lovász and Schrijver.We introduce a new property of graphs called “q-state Potts uniqueness” and relate it to chromatic and Tutte uniqueness, and also to “chromatic–flow uniqueness”, recently studied by Duan, Wu and Yu.     

Projection Pursuit Indexes Based on Orthonormal Function Expansions

Abstract

Projection pursuit describes a procedure for searching high-dimensional data for “interesting” low-dimensional projections via the optimization of a criterion function called the projection pursuit index. By empirically examining the optimization process for several projection pursuit indexes, we observed differences in the types of structure that maximized each index. We were especially curious about differences between two indexes based on expansions in terms of orthogonal polynomials, the Legendre index, and the Hermite index. Being fast to compute, these indexes are ideally suited for dynamic graphics implementations.

Both Legendre and Hermite indexes are weighted L ² distances between the density of the projected data and a standard normal density. A general form for this type of index is introduced that encompasses both indexes. The form clarifies the effects of the weight function on the index's sensitivity to differences from normality, highlighting some conceptual problems with the Legendre and Hermite indexes. A new index, called the Natural Hermite index, which alleviates some of these problems, is introduced.

A polynomial expansion of the data density reduces the form of the index to a sum of squares of the coefficients used in the expansion. This drew our attention to examining these coefficients as indexes in their own right. We found that the first two coefficients, and the lowest-order indexes produced by them, are the most useful ones for practical data exploration because they respond to structure that can be analytically identified, and because they have “long-sighted” vision that enables them to “see” large structure from a distance. Complementing this low-order behavior, the higher-order indexes are “short-sighted.” They are able to see intricate structure, but only when they are close to it.

We also show some practical use of projection pursuit using the polynomial indexes, including a discovery of previously unseen structure in a set of telephone usage data, and two cautionary examples which illustrate that structure found is not always meaningful.     

Exploring Graduate‐Level Scientists' Participation in a Sustained K‐12 Teaching Collaboration

To increase scientific literacy in America, the National Science Foundation has implemented the Graduate Teaching Fellows in K‐12 Initiative (GK‐12). In these programs graduate level scientists, known as Graduate Teaching Fellows (GTFs), act as resources for science teachers (Partner Teachers). This research examines the influence of participation in this program on the GTFs involved in one GK‐12 sustained collaboration that emphasizes the codevelopment and co‐implementation of hands‐on, inquiry‐based activities. Qualitative data in the form of interviews and observations were collected from a single cohort of Partner Teachers, GTFs, and the academic advisors of the GTFs for one academic year. The GTFs identified three benefits they received from working in the GK‐12 program: enhanced understanding of science content, fuller understanding of the complexities of teaching science, and understanding of inquiry‐based science teaching and its value. At the same time the GTFs experienced tensions related to differing views of science and science teaching held by the GTFs and the Partner Teachers. The GTFs also experienced professional risks in at least two ways due to their involvement in this program. One risk was the result of the time commitment required to be a GTF. The other was the result of what was termed competing foci. Illustrated in this case by the difficulties encountered as GTFs attempted to learn how to teach in a secondary science classroom while still learning to become research scientists.     

An analytic method in general relativity

An analytic method, which Wu called the “Bochner technique,” has been used for fifty years to describe global Riemannian and Kähler geometries. We use this method to describe conformally Killing vector fields and harmonic timelike vector fields on a Lorentzian manifold and to study hydrodynamic models of the Universe, the existence of closed spacelike sections, and the possibility of fibering Lorentzian manifolds.     

Undergraduate computational science and engineering education: A SIAM Working Group Report

Computational Science and Engineering (CSE) is a rapidly growing multidisciplinary area with connections ot the sciences, engineering, mathematics and computer science. CSE is a legitimate and important academic enterprise, even if it is yet to be formally recognized by a number of institutions. The undergraduate arena is the most important segment of the educational pipeline, since it prepares teachers for the high school environment, invigorates students to pursue graduate studies in cutting edgetechnical fields, and produces a vast number of future employees for industry and the “knowledge-based” economy. Thus it is critical that CSE curricula and programs are a viable option for every undergraduate student. This presentation introduces the SIAM Working Group report on undergraduate CSE education. The particular focus here is on background and the varying nature of programs. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Examination-Type Preferences of College Science Students and Their Faculty in Israel and USA: A Comparative Study

The science-examination preferences of college science students and their science faculty were surveyed, using the TOPE questionnaire at a teacher training and a community college in Israel and the U.S., respectively. The results obtained in the two countries were “intrally” and “interly” compared, in total and by gender, in terms of significant/no significant differences in the preferences made and the reasons provided by the students and faculty for their ranking. The findings suggest that: (a) college science students prefer mostly, the Israelis more so than the Americans, the nonconventional, written exams in which time is unlimited and any materials are allowed; (b) American college science students prefer the traditional class science examination (G) significantly more than their Israeli counterparts; (c) the preference of higher order cognitive skills (HOCS)-oriented exams (B. I and H) is significantly higher for female science students in Israel compared with no gender difference concerning the preferred examinations in the US, and rejection of oral examinations by all in both countries, significantly more by female students; and (d) there exists a significant gap between the preferred type of examinations of science students and their faculty in both countries. In view of the HOCS-orientation and the goal of conceptual understanding in current reforms of science education worldwide, the consonance between these curriculum objectives and examination practices is advocated. This, in turn, requires that provisions be made to lessen the gap between science teachers and their students' examination type preferences for better science learning to occur.