A VISION FOR DATA SCIENCE AT THE CLAREMONT COLLEGES June 2018 (orig drafted 10 May 2018) Contributing Authors*†: Cultivating Capacity Within Our Faculties Defining Curricular Pathways Jeanine Finn, LIB Findley Finseth, Keck Science Jeho Park, HMC Shana Levin, CMC Greg Reardon, KGI Jemma Lorenat, PTZ Hovig Tchalian, CGU Julie Medero, HMC Chris Towse, SCR Roberto Pedace, SCR Charge to Writing Teams: Working under the broad umbrellas of “cultivating capacity within our faculties” and “defining curricular pathways,” foci identified by the Academic Deans Committee (ADC), two writing teams met May 10, 2018 to draft preliminary documents articulating coherent visions to inform professional development and a data science curricular framework, respectively, to be implemented across the consortium Drawing on key ideas‡ surfaced in the Fall 2017 interviews (50 1:1 and small group) and charrettes (48 participants)—ideas that were also perceived by the deans and presidents as resonant with their institutional interests—each team authored a five-page document concretizing how a compelling presence for data science at the 7Cs could be realized This unified vision statement is intended to be shared with faculty and staff in the Claremont data science community, students, colleagues at other institutions, corporate partners, and potential funders for reactions and feedback Working Definition of Data Science: Data Science: The creation, application, and critique of tools and processes that enable users to extract and communicate meaningful insights from data drawn from across the disciplines Data scientists make data useful * Editor: Christine Costanza, Admin Coordinator, OCAC † Teams were convened with the following governing parameters: that each college be represented (as well as Keck Science and the Library); that no more than one person from a given institution sit on a team; and that teams benefit from a range of disciplinary perspectives while at the same time being small enough to ensure productive engagement (The Pomona representative was unable to attend the day-of.) ‡ See Appendix A for a list of specific initiatives/proposals gathered through OCAC’s process assigned each team Data Science Visioning at The Claremont Colleges 6-08-2018 Page I Vision statement A Grounded in the Liberal Arts At the highest level, our vision sees data science literacy as a core and common competency across each the colleges In 2018, it seems less a matter of “if” The Claremont Colleges will develop its data science capacity and more a matter of “how.” The teaching and learning happening across our campuses are deeply rooted in the traditional culture of a liberal arts college, while also benefiting from the dynamism of 21st-century interdisciplinary collaborations and the rich higher education resources of Southern California The Claremont Colleges are poised to create a model for data science education that builds and applies tools and methods into a well-grounded, contextually-nuanced, and critically-examined course of study in the liberal arts tradition Given a unique consortial environment built upon liberal arts education, faculty from a wide variety of disciplines should be invited and included in developing new resources and leveraging already existing resources for incorporating data science in their teaching and research The interdisciplinary nature of data science can well be applied and fit to our unique educational environment if we encourage inclusive activities among faculty Particular programs might typically start at the level of an individual college or department that has a particular educational or research need that might be solved by the effort Alternatively, a group of faculty might have a particular affiliative interest in a data science topic, method or application A data science curriculum, construed below, should offer students both breadth and depth of study in the context of the liberal arts; it will expose students to critical considerations in ethics, the application of data analytic techniques to address pressing social problems and frame policy debates, and the communication of findings within both public and academic domains B A Data Science Curriculum We see research and teaching as an interconnected loop for developing faculty capacity for data science expertise New research approaches leveraging data science could inform curriculum development, while classroom teaching and learning will inspire new research trajectories across disciplines and colleges Faculty will be able to modify or expand existing courses to include a data science component This would be supported by an intercollegiate framework of support for learning new applications This framework would similarly support individual and programmatic research agendas, leading to innovations our programs would be known for We posit four essential elements of the data science curriculum: an introductory course in programming, a foundations of data science course, a course in introductory statistics and elective courses in the application of data science techniques within specific knowledge domains Breadth of application could span across disciplines, types of data or data analytic tools (e.g., qualitative vs quantitative, discrete vs continuous, parametric vs nonparametric) Depth of study is reflected in the sequencing of courses from introductory programming, data science and statistics to Data Science Visioning at The Claremont Colleges 6-08-2018 Page intermediate and advanced elective courses in the application of programming, statistical, and data science tools within specific knowledge domains The data science curriculum will fit into current institutional structures, such as a minor at Scripps, Pitzer and Pomona, certificate at HMC and CGU, and sequence at CMC The learning outcomes of the curriculum (see below and Appendix B) are to ensure that students will be able to create, apply, and critique tools and processes that enable users to extract and communicate meaningful insights from data drawn from across the disciplines In its entirety, through an institution-specific number of courses drawn from the four elements identified in the previous paragraph, the curriculum should cover six main learning outcomes (SLOs) derived from our working definition for data science: The (1) creation, (2) application, and (3) critique of tools and processes that enable users to (4) extract and (5) communicate meaningful insights from (6) data drawn from across the disciplines (See Appendix B for an explanation of the SLOs.) Three foundational courses (an introductory computer science course, an introductory statistics course, and a Foundations of Data Science course) would cover SLOs through and (two or three) cross-disciplinary applications courses would collectively cover aspects of SLOs through again ● Introductory computing course (objective 1) ● Introduction to Statistics (objectives 2, 3, 4, 5) ● A new “Foundations of Data Science” Course (objectives 1-6) ● Applications, including a clinic/research/capstone experience C Faculty Development For developing capacity within our faculty we are envisioning a primary focus on growing data science expertise (fluency with tools, theories, and research methodologies) within existing disciplines in the colleges We see this as the “horizontal” component that will build on existing foundations of curriculum and research specialities within our departments As this horizontal capacity develops, we envision a growing vertical component, as graduate/undergraduate/ institutional data science initiatives and expertise mature at The Claremont Colleges Over time, data science as a specific area of inquiry could establish itself more formally in our institutions, finding a disciplinary home in computer science, science and technology studies, economics, or any number of areas depending on the college and faculty strengths Data Science Visioning at The Claremont Colleges 6-08-2018 Page The emerging needs for faculty capacity in data science at The Claremont Colleges will be informed by feedback from ongoing vehicles for collaboration among faculty that may center around particular resources (e.g., high performance computing resources) or particular research methodologies (e.g., the text analysis working group currently organized through the Digital Humanities center) Student interest and feedback will be captured through declarations of relevant majors, minors, sequences, etc; course enrollments, and senior thesis topics There are also a number of potential synergies with efforts at University of California campuses and other nearby institutions that could inform the direction of data science at The Claremont Colleges Data science workshops and professional development opportunities should also be regularly provided with incentives (e.g., stipends and funds) for faculty activities for attending data science workshops and faculty development events D Sustainability One key goal of the data science vision proposed here is to generate a sustainable source of funding An essential component of any data science initiative proposed will begin with this end in mind Specifically, how will the data science initiative generate attention and interest that, in turn, results in sustainable demand for the benefits offered and subsequent funding to sustain the program once seed funding is exhausted This addresses the general concern among The Claremont Colleges’ leadership surrounding whether any proposed data science program, particularly one that is likely to require substantial allocation of faculty talent and funding resources, is able to demonstrate its value in terms of either directly enhancing the student experience or building visibility for the colleges that translate into a largely enhanced reputation We also recommend holding faculty forums and meetings to generate ideas tapping into existing college resources with sustainable and permanent funding models II Implementation: What work would be needed to realize this vision? A Cultivate Capacity First, achieving centrality of purpose in data science programs across the 7Cs will require establishing an ambitious core or home for the data initiative, one that is adequately purposed, supported and funded Second, innovative ideas must be encouraged, recognized and, where successful, allowed to scale into sustainable centers of excellence, either in teaching or research, residing within or across particular colleges The liberal arts approach the Colleges are known for sets us apart and provides a distinguishing characteristic Data Science Visioning at The Claremont Colleges 6-08-2018 Page B Emphasize the Liberal Arts It is critical that the initiative start with the existing assets in our faculty In order to enhance and encourage the inclusion of data science (ideas and skills) into courses across the curriculum, we would need to allow for the training of existing faculty in particular aspects of data science Mini-courses specifically designed for faculty would need to be developed and offered regularly (Examples might be programming courses in Python using Jupyter notebook environment or courses in the use of R, SPSS or other statistical software.) The primary goal of offering such courses is to enable faculty to redesign their course offerings in a way that addresses data science questions within their disciplines, allowing development across various ranges of tenure: early, current and future In addition, these courses will bring together faculty from myriad disciplines who have interests in data science Follow-up project-based courses and/or reading groups may develop out of the initial introductory courses, as well as potential research collaborations Opportunities to co-teach courses in Data Science may also develop out of these interactions C Incentivize Innovation In order to supervise the Data Science programs and coordinate the Data Science courses (including managing the approval of such courses) a Director of the 7C Data Science Initiative (DSI) would need to be named This position would rotate through the colleges and would coordinate with the Center for Teaching and Learning in offering the faculty mini-courses The Director also would work with the registrars to cross list courses as DATA D Facilitate Collaboration and Provide Resources Existing modes of intercampus communication are largely currently limited to email and listserv distributions, but the scalability of such a primitive platform for building one-to-one or even one-to-many affiliations among approximately 10,000 students and 800 faculty is not practical One solution to this problem would be to develop a single center, that would economize the extremely large volume of interactions that would be necessary for the data science initiative to develop and sustain The DSI could serve as a type of incubator, facilitating, for instance, the startup of cross-college affiliative groups with aims for developing fundable research initiatives, with matching student and faculty mentors for research or scholarly endeavors of common interest (e.g Pomona GPUs), or for pursuing grant opportunities Data science brings with it some unique needs that will require ownership and clear lines of responsibility across colleges and departments These needs could include sites for curated data education resources, research tools (such as software licensing, collaborative platforms, and data repositories), and regular oversight and coordination of faculty training opportunities Data Science Visioning at The Claremont Colleges 6-08-2018 Page III Cost: What budget resources would be needed to realize this vision? A Funding ■ for course development for faculty to adapt existing courses to satisfy Data Science requirements, or to develop new such courses ■ To seed interdisciplinary, collaborative, and innovative research projects incorporating data science ■ for creating, running, and attending faculty mini-courses Claremont faculty to be compensated with course development funds and pay for creating and running mini-courses Stipends for faculty who attend ■ for TA positions for graduate students B Course release… ■ for (rotating) Director of DSI C Personnel… ■ to teach the Data Science foundational course ■ to serve as coordinator/s: with instructors (how to incorporate SLOs within their courses), with departments (teaching a foundational course in DS would take an FTE away from a department), with registrars (DS label in the portal), for assessment (of SLOs and of the program structure) ■ to comprise a permanent cross-campus coordinating committee D Centralized resources for computational analyses, data storage, data management, data licensing for classes and research Data Science Visioning at The Claremont Colleges 6-08-2018 Page APPENDIX A Ideas to Cohere Into a Compelling Presence for Data Science at the 7Cs Cultivating Capacity Within our Faculties: Establish brown bags/works-in-progress; Distribute a roster of interested people, their data, their tools, their questions; Host informal drop-in events: Research Salon, "Open Mic Night," happy hour; co-teach courses; Host an annual research conference; Offer training for faculty (e.g., skills seminars, one-week summer intensive); Build out a superstructure (and admin support) for data storage and processing (e.g., computer cluster, HIPAA-compliant storage); Establish TA opportunities for CGU students (e.g., grad TAs could bring DS modules into your existing course) Defining Curricular Pathways: Cross-list courses as DATA; Launch a course redesign initiative to bring data science into existing classes and labs (e.g., modules); Articulate a coherent DS curriculum that advances beyond the introductory level (e.g., major, minor, microcredentials, certificate); Create an Intro to DS course that can serve all Claremont students; Offer an intensive academic program students could complete during a summer Data Science Visioning at The Claremont Colleges 6-08-2018 Page APPENDIX B Student Learning Outcomes Creation of tools and processes: a Developing computational thinking - articulating solutions to questions/problems in code b Articulating a solution to a machine in a way that also makes sense to other humans c Un-blackboxing computers (both as software and as a physical machine) d Understanding what machines can and cannot e Build models of data: optimization, supervised and unsupervised machine learning f Data storage (e.g., databases) g Import, explore, clean, manipulate data Application of tools and processes: a Handle data to answer specific questions in real-world contexts b Apply models of data to predict and forecast c Working from clean data to answer given questions (hypothesis driven) d Finding data to answer student generated questions (discovery based) e Using arguments drawn from data to develop courses of action Critique tools and processes: a Ethics of data and data analysis b Un-blackboxing results c The cost and process of data storage d Bias in data collection, analysis, distribution e Determining who has access to what data (ownership and privacy) f History of how data has been gathered, analyzed, communicated across cultures and over time Extract meaningful insights from data: a Bayesian and traditional statistics b Quantitative and qualitative c Continuous and discrete (categorical) d Big and less big e Parametric and non-parametric Communicate meaningful insights from data: a Integrate quantitative results with hypotheses and results b Researching background literature and placing findings in a broader context within and beyond a field of inquiry c Understanding one’s audience and what they want to know from the data d How the same data can be communicated in different and even contradictory ways (telling stories with data) e Developing and interpreting practices of visualizing and representing results .From across the disciplines a Context of the liberal arts b Emphasize breadth c Differing disciplinary-specific guidelines (e.g what counts as “significant,” what counts as “big,” what counts as “strong”)