This CSUMS project, NCSU CUSP, prepares students to engage in a significant research experience, and to be fluent in the languages of computing, mathematics, and statistics. The investigator and his colleagues provide inquiry-guided learning to train undergraduates in exciting new approaches to problems involving massive datasets or computationally intensive methods. The program couples extensive coursework in computing for contemporary statistical analysis with a practicum and research lab focusing on an area of application such as drug discovery, pattern recognition, statistical genetics, data assimilation, or financial risk. The practicum and lab provide scientific background, an overview of statistical and computational approaches, training and feedback on teamwork and communication skills, and a significant mentored research experience. Faculty members from interdisciplinary teams conducting research in the areas of data mining, geophysical and environmental data assimilation, statistical genetics and bioinformatics, Bayesian hierarchical modeling, financial risk modeling and time series analysis serve as NCSU CUSP mentors and as co-mentors for students from neighboring Meredith College. The NCSU CUSP meets four specific objectives. It (1) prepares students to take advantage of computing advances and make sophisticated computing an integral part of the statistical and mathematical methodology curriculum and research experience; (2) improves students' non-technical skills, including public speaking and written communication, working in teams, and ethical reasoning, (3) provides the research experience to apply initiative and creativity in developing statistical and computing approaches to interdisciplinary scientific problems; and (4) prepares and motivates a diverse pool of highly qualified students to pursue interdisciplinary graduate studies in the mathematical and computational sciences.
Aided by rapid advances in technology, massive amounts of new data are generated daily in many scientific disciplines and the volumes are growing at a rate unprecedented in human history. For the US to remain competitive and innovative, a diverse pool of researchers trained in novel and powerful techniques is critically needed to illustrate, model, and analyze these large-sized, high-dimensional, and nonlinearly-structured data. Building on resources of one of the country's largest statistics departments, NCSU CUSP becomes one of the first computationally intensive statistics programs for undergraduates in the nation. This project leads to development of new computationally intensive courses and interdisciplinary courses, which will have a long term impact. The project is also committed at the outset to increasing diversity in the emerging field of computational statistics. NCSU CUSP increases awareness of statistical science among mathematics majors and faculty, fosters greater collaboration between interdisciplinary programs, and encourages a diverse pool of well-prepared students to pursue graduate studies in quantitative sciences. Training in communication skills helps develop graduates who can bring scientific research results to the public and policy makers. NCSU CUSP methods and results are widely publicized to the mathematical science and computational communities through regional and national conferences, participation in workshops, a CUSP website, and published articles. The project is supported by the MPS Division of Mathematical Sciences, the MPS Office of Multidisciplinary Activities, and the EHR Division of Undergraduate Education.
With rapid advances in technology, massive amounts of new data are generated daily in many scientific disciplines and the volumes are growing at a rate unprecedented in human history. For the USA to remain competitive and innovative, a diverse pool of researchers trained in novel and powerful techniques is critically needed to illustrate and analyze large and complex data. The North Carolina State University Computation for Undergraduates in Statistics Program (NCSU-CUSP) was designed to engage talented undergraduate students in research at an early stage of their careers. The primary objectives were to give students a practical research experience to solidify the concepts developed in their coursework and to provide top students with the information and motivation needed to pursue graduate school in a STEM field. In each year of the project, a cohort of 5-8 students were provided classroom training on topics including manipulating large datasets, advanced computational techniques, and statistical methods appropriate for their research projects. Students were then paired with scientific collaborators from diverse fields such as genetics, atmospheric science, and ecology. The students conducted research for an entire academic year and presented their work at top undergraduate statistics conferences. Intellectual merit: The project trained over 30 students over the six years, including many students from underrepresented groups. The students produced high quality research. As an example of the nature and diversity of the research, consider the work by the three student groups in the final (2014) cohort. The first group studied measurement error in blood alcohol content (BAC) readings and produced a look-up table giving the likelihood that the true BAC level exceeds the legal limit based on two noisy readings. The second group investigated the effectiveness of intervention strategies to curb the spread of white nose syndrome in bats using Monte Carlo simulation and systems dynamics models of disease spread. The third group compared the efficiency of mathematical shortest-path algorithms with the paths obtained using slime mold populations grown in the lab. Over the course of the program, the work resulted in several peer-reviewed publications and the students presented their work at regional, national, and international conferences including meetings in Seattle, Montreal, and Boston. Broader impacts: The primary impact of the program was to prepare a cadre of young statisticians eager and prepared to pursue research careers. The program was quite successful in this regard. Many students pursued master's degrees in fields such as statistics, biostatistics, analytics, and math education. Students also entered PhD statistics programs at North Carolina State University, The Ohio State University, and Columbia University. Other students entered the work force directly after finishing the bachelor's degree to work for companies such as Bank of America and the Environmental Protection Agency. These students left North Carolina State University equipped with computational skills and practical experience required to thrive in a technical and multidisciplinary environment. NCSU-CUSP has increased the awareness of statistical science among minority students and faculty, fostered greater collaboration between departments, and encouraged a diverse pool of well-prepared students to pursue graduate studies in quantitative sciences. CUSP is a model for how programs to improved undergraduate research should work. Many programs implement Research Experiences for Undergraduates (REUs). CUSP goes beyond this typical model by incorporating a cohort structure that provides a built-in support mechanism for participants. It also supplements traditional research activities with new courses that train students in methods that they can apply immediately. This cohort structure combined with curricular transformation creates a model that can make undergraduate research work elsewhere. We firmly believe that CUSP is a program that is making a real difference among the students at NCSU and can serve as a model for real transformation at other institutions.
