Content area
Career choices are shaped by students’ experiences, knowledge, and skill sets across time, reflecting not only disciplinary interests but also exposure to evolving fields such as data science (DSC). Despite a surge in interest and enrollment in data science degrees, the United States faces a growing demand for data literacy across multiple sectors. Online learning environments have become entry points for students’ initial engagement with DSC, offering accessibility and supporting workforce needs. Nevertheless, the interdisciplinary essence of DSC means that clear career paths remain ambiguous, especially for those applying DSC knowledge within various disciplines. While national data sources provide valuable overviews of degree distributions, more granular analysis at the course level is warranted to understand nuanced student trajectories. Project-based online learning, though proven valuable in in-person settings, remains underexplored in online DSC education. This study employs curriculum analytics and Sankey diagram visualizations to investigate course enrollment patterns and career trajectories among students after enrolling in an introductory online project-based DSC course. We built a longitudinal dataset by following 35 students between Fall 2022 and Spring 2024, tracking their subsequent course enrollments over time. Demographic and academic data were sourced from institutional enrollment records, allowing subgroup analysis based on major, gender, race, first-generation status, and achievement. Our exploratory analysis reveals patterns indicating that continued DSC course enrollment appears prevalent among nonwhite, male, STEM-major, and academically proficient students, whereas first-generation students exhibit no persistence. We illustrate how Sankey diagrams, though not establishing causality, provide actionable insights for program and curriculum development in DSC education.
Introduction
Rawlings et al. (2023) pose that career choices “entail notions of past, present, and future” (p. 259). Therefore, exploring students’ career choices could give us insight into their prospective job placement. For example, exploring the patterns that emerge from visualizing students’ curricular decisions after enrolling in an elective course and their relationship with students envisioning their job prospects and contributions to society. Holland and Lochicotte (2007) argue that students’ career choices are influenced by their professional identity and social contexts. In this study, career choices are understood as students’ course enrollment decisions, particularly in elective contexts where they have greater autonomy over their academic pathways. Course enrollment behaviors serve as meaningful indicators of career exploration, as they represent students’ cognitive engagement with planning for program completion (Tatel et al., 2022) and reflect individual preferences about future directions (Pardos & Nam, 2020). This approach allows us to examine how students navigate their “circuitous journeys” toward developing occupational identity (Advance CTE, 2021). Building on this understanding, students may unconsciously limit their career options to those they’ve been exposed to through their social environment, potentially hindering their exploration of alternative opportunities. In other words, the phenomenon of following the career path that your parents did. Therefore, it is urgent to examine students’ course choices that might expose them to various career pathways and help us identify hypotheses about students’ future workforce experiences. Despite the latest interest surge in data science (DSC) degrees (Coffey, 2024; Pierson, 2023), the US needs to meet the workforce demands for data literacy not only in traditionally data-oriented fields like finance and business intelligence but also across different industry sectors. This is especially true given the interdisciplinary applications of DSC that range from higher education to transportation to commerce (Coffey, 2024). Therefore, it is necessary to create undergraduate learning experiences to attract, retain, support, and empower learners with multiple disciplinary interests in DSC. In response to such need, the All-campus Data science and AI Project-based Teaching and learning (ADAPT) course model emerged to support DSC instruction and workforce development (Data Science and Artifical Intelligence Academy, n.d.). Through the courses designed using the ADAPT course model, students from various disciplinary backgrounds engage in DSC projects in a way that affirms their primary career choice while developing skills that prepare them for on-the-job tasks.
Yet, for many, the first approximation to DSC is through online delivered materials. Before DSC programs’ growth and enrollment increase in higher education (Coffey, 2024), educational vendors already offered numerous non-credit online learning options for those interested in DSC. Yet, most vendors’ curriculum focuses on data practices (e.g., data wrangling), leaving out the application of DSC in various contexts. Therefore, it is essential to understand how online instruction in DSC could be used in for other DSC skills apart from delivering only technical content. Integrating literature on human capital in data science, online project-based learning, and curriculum analytics, we explore students’ course choices based on their course-taking behaviors after enrolling in an online-project-based learning DSC course.
Literature review
Course choices as indicators of career choices in data science
Career choices mirror the structure of students’ knowledge, experiences, disciplinary interests, and skill sets. Hence, course choices could provide a tentative prediction of students’ career prospects and their intended contributions to society. Students invest in their education to acquire appropriate human and cultural capital that enable them to signal their fit for their desired job and chosen career. Education also provides students a space to engage with the discipline they wish to enter and visualize themselves working in that capacity. As such, majors in higher education are designed to inculcate students into the disciplines they want to pursue by offering industry and discipline-specific training in required courses (Fiorini et al., 2023). However, this highly structured training might prevent students from exploring alternative career paths through their education. This is especially true for “gatekeeping” (Gasiewski et al., 2012) courses that could deter students from considering careers in DSC.
The field of DSC builds upon mathematics, statistics, and computer science to provide multi- and interdisciplinary applications (Friedman & Beasley, 2024). Its convoluted origins have drawn students from all interests to enroll in DSC programs and courses, sparking a popularity in DSC majors that could be described as “the new girl on the block” (Findley et al., 2025, p. 8). Figures from the National Center of Educational Statistics (NCES) signal that the growth in DSC programs is no less than dramatic, with an 81% increase in master’s degrees awarded from 2020 to 2023 (Pierson, 2024). For students to enter graduate school for a DSC program, it is necessary for them to explore the field at the early stages of their careers. In a qualitative study, Findley et al. (2025) found that undergraduate students choose DSC as a starting point for their undergraduate major as a vocation, but others see DSC as an interdisciplinary trade that leaves room for their true passions (e.g., dance, baseball) while expanding their future job opportunities.
With regard to widespread participation in DSC, scholars have long debated the absence of specific subpopulations in STEM disciplines (National Science Board, 2024) and, by extension, in fields contributing to DSC workforce. Of particular interest are demographic characteristics like race and gender. In terms of race, the American Statistical Association reported (Pierson, 2024), based on the 2023 figures from NCES, that the percentage of degrees earned in statistics, data science, and analytics of students who identify as white make the 55 to 62%. Regarding other races, the numbers are within the single digits (4–8% for Hispanics or Latinos and 2–3% for Black or African Americans) and reach zero for certain subgroups (0% for Alaska Native and 0–1% for Native Hawaiian or Pacific Islander). On the question of gender, Pierson (2023) reports that women make up 42–43% of the bachelor’s and master’s degrees earned in statistics based on 2022 NCES data, depicting a balance among men and women.
Yet, being DSC an interdisciplinary endeavor, it is still unclear what the potential career path in DSC is for those who are exploring DSC applications in their disciplines. In other words, those in the humanities and arts or social sciences. Despite NCES offering a top-down, robust account of the distribution of degrees awarded based on demographic characteristics, we argue that more granularity is needed to examine subgroups, for example, at the course-choice level. Without examining what courses students take during their careers, we are missing all those who enter DSC careers without degrees in either DSC or related fields. When studying the information technology labor market in a computer science online master’s degree, which is akin to DSC, Ruthotto et al. (2021) argue that as the industry diversity of graduate program applicants increases, students start graduate programs without job experience or prior degrees in fields associated with DSC. Therefore, it is crucial to examine not only graduation but also students’ choices within their graduate programs over time, in other words, their careers as students.
We argue that course choices are indicators of career interests, especially in elective courses. For example, Tatel et al. (2022) found that course enrollment behaviors serve as indicators of cognitive engagement and completion, recommending the relaxation of program requirements. Tatel et al. (2022) note that learners are constrained in course choices by various circumstances including prerequisites/program requirements, curricular structure, and course availability. Yet, we argue that elective courses provide students with greater agency in their academic decisions. The patterns that emerge after students take elective courses in DSC have not yet been explored. Furthermore, elective courses help us investigate those who would not usually take DSC courses, for example, non-STEM majors. The Career Readiness Framework (Advance CTE: State Leaders Connecting to Learning to Work, 2021) emphasizes that “learners often take circuitous journeys” (p. 2) while understanding their career options and the best path to attain them, also known as occupational identity. Pardos and Joo Hun Nam (2020) argued that “it is ultimately the student’s individual preferences […] that decide which courses will be taken” (p. 20). As such, the granularity of course choices allows us to observe the nuanced decisions that students make in their paths through higher education, providing insight into their potential careers.
Online project-based learning for data science education
Even accounting for the post-pandemic decline in online learning enrollments, online education growth is projected to continue (Coffey, 2024). Such growth could be particularly true in the case of DSC education, where higher education institutions have responded to the demand for DSC programs (Irizarry, 2020; Pierson, 2023). Online learning in DSC and affiliated fields (e.g., statistics, computer science) has become essential to meet the needs of the modern workforce and increase access to STEM disciplines (Ruthotto et al., 2021). The flexibility and accessibility of online programs are crucial for training individuals in this rapidly evolving field and providing access to those who cannot attend brick-and-mortar higher education institutions (Deming et al., 2015; Goodman et al., 2019). It is necessary for DSC programs nationwide to survive the hype of their name (Findley et al., 2025; Irizarry, 2020) and guarantee program sustainability for their institutions (Parker et al., 2021). We argue that online learning is a promising option that can help solidify higher education DSC programs with the potential to make data science education more accessible.
DSC is an inherently interdisciplinary field where students must collaborate across knowledge domains to address field-generated problems. Therefore, we posit that lecture-like instructional approaches in online learning are insufficient to provide a high-quality DSC education experience that responds to the workforce’s needs. One way to address this issue is through online project-based learning as an instructional approach to foster knowledge-construction processes by engaging in real-world projects. Meta-analytic evidence from 66 studies shows that project-based learning instruction has a large impact on students’ outcomes and academic achievement (SMD = 0.65) and a moderate effect on students’ emotional attitudes (SMD = 0.389) and thinking skills (SMD = 0.386) (Zhang & Ma, 2023). Regarding online learning, Koh et al. (2010) pioneered the exploration of project-based learning for knowledge construction in asynchronous instruction through discourse analysis of online learners’ discussion board text-based interactions. They found that students showed higher levels of knowledge construction in weeks with project-based tasks compared to weeks with non-project based tasks. Also examining discussion boards, Guo et al. (2021) found that online learners’ levels of effectiveness and knowledge exploration were predictors of academic performance in project-based learning. Guo et al.’s (2021) work evidences that online-project-based learning is an optimal alternative to serving DSC students at a distance.
Despite promoting project-based learning in traditionally in-person undergraduate DSC courses (Donoghue et al., 2021), a dearth of research has explored online-project-based learning in DSC education. As a pandemic-driven response, only Sakamaki et al. (2022) focused on project-based learning data science delivered via synchronous online learning, in other words, videoconferencing software (e.g., Zoom). However, given the timing of data collection in 2020, it is safe to assume that instruction was focused on Emergency Remote Teaching (ERT) (Hodges et al., 2020) rather than instruction designed to be delivered online. Therefore, it is of utmost need to investigate DSC education experiences that intended to be taught online and the extent to which they support students’ career development.
Curriculum analytics to investigate data science course choices
Curriculum analytics involves the holistic analysis of students’ learning trajectories by revealing relationships within their learning pathways and course choices (Dawson & Huball, 2014). Curriculum analytics serves to identify potential bottlenecks that could become gatekeeping courses by contrasting the learned curriculum (i.e., students’ actual course-taking behaviors) with the planned curriculum (i.e., degree program’s intended course sequence) (Mashile et al., 2023). Simon De Blas et al. (2021) clarify that curriculum analytics could take two forms: one, as graphs that depend on prerequisite courses or, two, as flow graphs in which course restrictions are not known as a priority. Further, curriculum analytics serves to investigate whether students’ trajectories differ based on their background. For example, Fiorini et al. (2023) analyzed biology-related curricula at two institutions using a process mining methodology. Their work focused on understanding whether students’ actual pathways reflect degree programs’ intended progression through courses and whether subgroups of students differ in their course-taking behaviors. By leveraging students’ trace data in a process mining approach, they developed process maps evidencing that students’ actual course-taking behaviors are strikingly different from the degree program’s design. Their findings evidenced that the default course of study did not support all students because less than 50% followed it, “including students from traditionally privileged backgrounds” (Fiorini et al., 2023, p. 13). Thus, it is necessary to investigate the role of the courses that deviate from the curricula and how these courses can serve all students.
Descriptive techniques, like Sankey diagrams, have also been used to understand students’ success and enrollment because they demand basic or no programming skills that produce visualizations responding to high-level questions and signaling deeper analyses (e.g., Heileman et al., 2015; Oran et al., 2022; Rodriguez et al., 2016; Wang et al., 2019). For example, Heileman et al. (2015) used Sankey diagrams to condense big data in higher education. Specific applications of Sankey diagrams as a form of curriculum analytics centered on examining students’ paths in natural science departments (Oran et al., 2022) and physics majors (Rodriguez et al., 2016). By leveraging Sankey diagrams to discover students’ success in a natural science major, Oran et al. (2022) found that male students were leaving the university at a larger proportion than female students. Similarly, Rodriguez et al. (2016) found that 76% more women who attempted upper-level physics courses eventually received a physics degree compared to only 50% of men. Despite the initial work on curriculum analytics for STEM courses using Sankey diagrams, there is a dearth of research focused on curriculum analytics of DSC curricula. Furthermore, existing research focuses on a top-down approach focused on improving the curriculum rather than investigating students’ career pathways (e.g., Yoshia et al., 2022). We propose that curriculum analytics is a promising approach to investigate DSC career pathways, particularly in relation to students’ experience as online learners in project-based learning instruction.
This study
Being a data scientist was predicted to be one of the trendiest jobs of the 21st century (Davenport & Patil, 2012). The popularity of the vocation is reflected in calls for preparing the upcoming generations of data scientists to meet the workforce’s needs (Cassel & Topi, 2016). Nevertheless, higher education programs designed to prepare the next generation of data scientists can be burdened with insurmountable prerequisites, demanding a long-term commitment to a full degree in statistics or a related field. These barriers decrease the opportunities for undergraduates to explore DSC as a career choice, regardless of their declared major. As a response, the Data Science and Artificial Intelligence Academy (DSA) proposed elective non-prerequisite courses that expose undergraduate students to DSC instruction focused on training for on-the-job tasks using a project-based learning approach. Additionally, the DSA Academy offers a portion of its portfolio of DSC courses in an online teaching and learning format to improve reach across campus. Given the potential of online learning to make instruction more widely available (Deming et al., 2015; Goodman et al., 2019), this study explores the course-taking behaviors of students enrolled in an interdisciplinary online-project-based learning DSC course. Specifically, we investigate the course-taking trajectories of students who enrolled in a DSC elective using Sankey visualizations to offer descriptive insights for informing program decisions and suggest subsequent statistical analysis. To our knowledge, no previous work has examined the career paths of students interested in DSC within their preparation time during undergraduate instruction. Therefore, it is critical for the DSC education community to scrutinize the potential of leveraging online-project-based instruction to support workforce preparation in DSC. The following research questions guided our inquiry:
RQ1: In what ways can an online project-based learning DSC course foster students’ data science course choices as reflected by their course-taking behaviors?
RQ2: To what extent do subgroups of students differ based on course-taking behaviors after taking a project-based online course in DSC?
Our research questions are designed to examine course-taking behavior patterns and disciplinary engagement breadth rather than individual student trajectories or class enrollment sizes. Our focus on course-taking behaviors as indicators of career interest allows us to capture the full spectrum of interdisciplinary engagement following DSC exposure. Consequently, our analytical approach intentionally aggregates transition patterns across the student population, with individual students potentially contributing to multiple disciplinary pathways within each wave when they enroll in courses across different fields. This methodological alignment ensures that our visualization approach directly supports our research objectives of understanding how exposure to DSC influences students’ academic choices across disciplines.
Case study of using online project-based learning to foster students’ course choices in data science
To answer our research questions, we focused on a course on data science for social good (DSFSG) taught as part of the DSA Academy, whose efforts centered around workforce preparedness for DSC and AI. The DSFSG course focuses on addressing societal challenges through data science applications in alignment with the United Nations’ Sustainable Development Goals. This course is part of the DSA Academy. It uses the All-campus Data Science and AI Project-based Teaching and learning Model (ADAPT) to design and deliver instruction to train the next generations of interdisciplinary data scientists. The ADAPT model integrates three key components: project-based learning, common data science learning elements, and workforce preparedness. Under the first element, students are prepared for the future of work using end-of-course capstone projects or sequential assignments that build on each other. Specifically, the DSFSG online course used a project-based learning approach in which students were exposed to real-world data science applications and explored alternative career pathways(Sakamaki et al., 2022). The second element of the ADAPT model focuses on common data science learning elements that allow cross-disciplinary understanding and are grouped into three categories: data perspectives (e.g., data as information rather than truth), data practices (e.g., data wrangling and communication), and data discoveries (e.g., impact of data). As part of the second element, the technical applications in the DSFSG course spanned from machine learning to natural language processing to image recognition. The third and final element of the ADAPT model centers on workforce preparedness, in which students practice decision-making that resembles on-the-job requirements and the agency expected from independent professionals. For example, choosing the data to solve a problem, the best tools to use, and the correct answers to tackle. Altogether, the ADAPT model strives to provide data science training to all learners of the community regardless of their education level (i.e., undergraduate and graduate students), chosen majors, and status (i.e., faculty, staff, community members).
Methods
Research context and data
Data was collected from a university-wide data science and artificial intelligence education center at a public R1 Southern University. Data collection occurred over six consecutive semesters from Fall 2022 until Summer 2024. The units of analysis were the courses in which students enrolled after taking an online course about data science for social good (DSFSG). This study used institutional student records collected every semester by higher education institutions. Specifically, we built a longitudinal dataset that followed 35 consenting students who enrolled in the DSFSG course anytime between Fall 2022 and Spring 2024 and their subsequent course enrollment over time. In this dataset, four students took DSFSG in Fall 2022, seven in Spring 2023, 16 in Fall 2023, and eight in Spring 2024. Students were from different majors across campus. Data collection for this study was part of a larger project on DSC education and was approved by the Institutional Review Board of our institution.
Analytical approach
We used a curriculum analytics approach to examine the course choices of students enrolled in an interdisciplinary course on data science. Curriculum analytics is increasingly popular for understanding students’ pathways in higher education and stems from the complementary fields of Educational Data Mining (EDM) and Learning Analytics (LAK) (Fiorini et al., 2023). Researchers also use curriculum analytics to conduct longitudinal analysis, which allows them to follow students’ trajectories over time using visualizations. We used such an approach via Sankey diagram visualizations (Heileman et al., 2015). Sankey diagrams support research by converting sheer volumes of complex data sets “into intuitive visual graphics that document student progress longitudinally” (Heileman et al., 2015, p. 30). We used Sankey diagrams because they produce solid descriptions with big data and small samples. Unlike other visualization approaches applied to big data in education, like dashboards, Sankey diagrams are flexible, tolerant of missing data, and require little to no programming experience.
Measures and procedures
Administrative records – Student & course level data
Students’ demographic characteristics were provided by the enrollment management office of our institution, which collects information such as academic plans, declared majors, racial and gender descriptors, first versus continuing-generation student status, and achievement. Using institutional data, we focused on making a subgroup analysis of students enrolled in the DSFSG course to find the extent to which this course could broaden the workforce development for subgroup populations and how it could support the course choices of all students in DSC and STEM disciplines. Course-level data included the course number, section, subject, and course title.
Transitions within courses
The longitudinal data was processed so that when a student first enrolled in the online course, DSFSG was considered the first wave of data. Given the various courses, we categorized students’ course choices based on the courses they enrolled in after taking DSGSG based on disciplines. We categorized the courses based on classifications from platforms used for seeking career advice, namely US News Disciplines and College Board (Coleman Fields, 2023; College Majors – BigFuture | College Board, n.d.). Although there is no clear-cut distinction on the boundaries among disciplines, we developed a list of seven categories to classify courses. Table 1 shows a categorization made in this study and a sample list of courses that fell under that category.
Although previous work has leveraged Sankey diagrams to understand student flow through academic programs and their path to graduation, we applied Sankey visualizations to explore students’ choice to enroll in courses from various disciplines after taking an interdisciplinary course on data science. Then, transitions among disciplines from one semester to the next one are accounted for as the flow represented by the Sankey diagram (e.g., DSFSG ◊ STEM ◊ Humanities). Our visualization methodology shifts from individual student tracking at the course level (Wave 1: DSFSG) to aggregated course transition analysis at the disciplinary level (subsequent waves), enabling us to identify broader enrollment patterns across academic domains. Further, numerical values and visual proportions of course transitions constitute the primary dataset for this exploratory study, enabling quantitative comparison of transition magnitudes across disciplinary categories. Data processing was done in R, and the visualization was done on the web-based platform Flourish Studio.
Table 1. Course categorization based on disciplines
Discipline Category | Sample Courses Topics |
|---|---|
STEM | Geographic information sciences, statistics, computer science, mathematics, physics, chemistry |
Business & Economics | Business management, economics, management innovation entrepreneurship, accounting |
Humanities & Arts | English, history, music, foreign languages |
Health & Medical Sciences | Health exercise studies (e.g., fitness, outdoor), nutrition |
Social Sciences | Psychology, anthropology, social work, sociology, Africana studies, political science |
Trades & Services | Landscape architecture |
Note. Categorization was made based on US News and the College Board’s list of reported disciplines. The sample list of course topics is not exhaustive
Results
Descriptive statistics
Table 2 shows the sociodemographic characteristics of the 35 students participating in this study. Wave one indicates the time when students first enrolled in DSFSG. The subsequent waves represent the following semesters after DSFSG until Spring 2024. We did not face attrition from wave one to wave two. However, the sample size decreased considerably after wave two. In this dataset, we find a larger proportion of females (60%) than males, white students (63%), students majoring in STEM degrees (60%), and continuing generation college students (68%).
Table 2. Sociodemographic characteristics of participants over time
Characteristics | Wave 1 n = 35 | Wave 2 n = 35 | Wave 3 n = 15 | Wave 4 n = 8 | Full Sample n = 35 | |||||
|---|---|---|---|---|---|---|---|---|---|---|
n | % | n | % | n | % | n | % | n | % | |
Gender | ||||||||||
Female | 21 | 60 | 21 | 60 | 11 | 73.3 | 7 | 87.5 | 21 | 60 |
Male | 14 | 40 | 14 | 40 | 4 | 26.6 | 1 | 12.5 | 14 | 40 |
Race | ||||||||||
White | 22 | 63 | 22 | 63 | 9 | 60 | 4 | 50 | 22 | 63 |
Nonwhite | 13 | 37 | 13 | 37 | 6 | 40 | 4 | 50 | 13 | 37 |
Major | ||||||||||
STEM | 21 | 60 | 21 | 60 | 11 | 73.3 | 5 | 62.5 | 21 | 60 |
Non-STEM | 14 | 40 | 14 | 40 | 4 | 26.6 | 3 | 37.5 | 14 | 40 |
First Generation | ||||||||||
Yes | 7 | 20 | 7 | 20 | 5 | 33 | 3 | 37.5 | 7 | 20 |
No | 24 | 68.5 | 24 | 68.5 | 7 | 47 | 2 | 25 | 24 | 68.4 |
Not reported | 4 | 11.5 | 4 | 11.5 | 3 | 20 | 3 | 37.5 | 4 | 11.5 |
RQ1: In what ways can an online project-based learning DSC course foster students’ data science course choices as reflected by their course-taking behaviors?
Sankey diagram interpretation
We used a Sankey visualization that aggregated students’ course-taking behaviors to explore students’ career paths in DSC after taking the DSFSG online project-based learning course. The diagram should be read from left to right, and the numbers on the top of each node indicate the wave in which the data was collected. In Fig. 1, the dark purple node in wave one represents when students first enrolled in DSFSG, and the subsequent nodes represent the disciplines of the courses in which they enrolled afterward. Wave 1 represents the specific starting point of our analysis (DSFSG enrollment), while subsequent waves aggregate courses by disciplinary categories to enable pattern analysis across broader academic domains. This progression from course-specific to discipline-aggregated analysis aligns with our research focus on interdisciplinary career exploration patterns.
Unlike previous uses of Sankey diagrams, links do not represent the number of students flowing through disciplines but rather the number of transitions made from the DSFSG course to courses in diverse disciplines. In other words, students’ course choices. Given that a single student enrolls in multiple classes of various disciplines, the numbers on top of the links are expected to add up to more than the sample size. The visualization intentionally captures all course transitions made by students, meaning that a single student enrolled in both STEM and Social Sciences courses will contribute to both disciplinary pathways. This approach allows us to observe the full breadth of interdisciplinary engagement following DSC exposure. Our approach intentionally aggregates course transition patterns rather than tracking individual student pathways across waves. Links represent the number of course transitions made from DSFSG to various disciplines, not the flow of individual students. Therefore, numbers represent course transition frequencies rather than student counts. The proportional sizing of links and nodes enables identification of the most prominent disciplinary transitions, which is essential for understanding course-taking pattern variations across academic domains.
General patterns after enrolling in DSFSG
In panel a of Fig. 1, one can observe the flow of student course-taking behavior after taking DSFSG, and in panel b, the highlighted counterintuitive course transitions (i.e., Humanities & Arts (HAS) and Social Sciences (SOCSCI)). We identified the DSC courses that had more than one student enrolled after wave one as “R for Biological Research,” “Introduction to R/Python for Data Science,” and “Data Wrangling and Web Scraping.” Again, in the case of DSC enrollment, we observed an expected need for data practices (e.g., data wrangling) while leaving room for data perspectives (e.g., biological research). It is worth mentioning that of the 13 DSC course offerings that students enrolled in after DSFSG, seven courses were delivered in an online education format. The increase in flow for course transitions in the HAS and SOCSCI is of interest here. Specifically, in panel b of Fig. 1 highlights the noticeable increase in waves one to two and two to three. Even though it was expected that transitions toward STEM courses were the most prevalent course choice, a sizable portion of transitions also occurred from STEM toward HAS and SOCSCI courses, as highlighted in panel b of Fig. 1. A closer inspection of which courses multiple students took in HAS and SOCSCI disciplines shows interdisciplinary course titles. Specifically, two courses were interesting: “Communication for Business and Management” and “Sociology of Medicine.”
[See PDF for image]
Fig. 1
Course-taking path after taking an online project-based-learning course in data science
Note: Links represent course transitions. Numbers indicate transition counts, not student counts. Since individuals enroll in multiple courses across disciplines within a wave, totals may exceed the sample size
Achievement
An exploration of students’ subsequent course-taking behaviors based on their achievement in the online project-based learning course in DSFSG showed counterintuitive results. Given that none of the students in the sample failed the DSFSG course, we categorized achievement between high and proficient. High achievement represented students who obtained grades A and A+ (n = 22), and proficient achievement those with grades A-, B, and Satisfactory (n = 13). High-achieving students did not show further course-taking behaviors in DSC courses as depicted in panel a of Fig. 2, whereas proficient achievement students continued taking DSC courses even two semesters later.
[See PDF for image]
Fig. 2
Course-taking path after taking the online project-based-learning course based on achievement
Note: Links represent course transitions. Numbers indicate transition counts, not student counts. Since individuals enroll in multiple courses across disciplines within a wave, totals may exceed the sample size
RQ2:. To what extent do subgroups of students differ based on course-taking behaviors after taking a project-based online course in DSC?
Gender
Subgroup analyses were conducted to understand how course-taking behaviors change based on students’ demographic characteristics. Regarding gender, we postulated that after taking DSGSF, female students’ (n = 21) course-taking flow will increase in STEM or DSC disciplines over time. Conversely, we expected male students to increase their course-taking behaviors in non-STEM disciplines. Figure 2 shows patterns consistent with some of our expectations. Specifically, panel a of Fig. 2 shows an increase in female students’ course transitions towards STEM courses in waves 3 and 4. The same phenomenon does not occur with DSC courses; female students do not continue enrolling in DSC courses after taking DSFSG. Panel b of Fig. 3 shows an small increase in male students’ course transitions towards Non-STEM disciplines (i.e., humanities & arts and social sciences) only in waves two to three. Further, the figures allow us to hypothesize that male students show more sustained DSC course enrollment patterns than female students following DSFSG. .
[See PDF for image]
Fig. 3
Course-taking path after taking the online project-based-learning course based on gender
Note: Links represent course transitions. Numbers indicate transition counts, not student counts. Since individuals enroll in multiple courses across disciplines within a wave, totals may exceed the sample size
STEM vs. non-STEM
Regarding majors, we observed similar patterns to those seen with gender. We initially speculated that the course-taking behaviors of STEM major students (n = 21) would increase toward enrolling in more non-STEM courses over time and vice versa, as non-STEM major students would drift toward STEM or DSC courses. The Sankey visualizations show patterns consistent only with the first speculation (Fig. 4, panel a). Most surprising is the increase in course transitions toward non-STEM courses of STEM major students in waves two to three. Further, STEM major students continued to enroll in DSC courses after their first enrollment in DSFSG in waves two to four. However, the enrollment flow of non-STEM majors in DSC courses stopped in wave two. Nevertheless, our descriptive results indicate that at least four course transitions occurred towards DSC courses and seven towards STEM courses, suggesting that DSFSG might open the breadth of disciplines available to non-STEM major students.
[See PDF for image]
Fig. 4
Course-taking path after taking the online project-based-learning course based on STEM major classification
Note: Links represent course transitions. Numbers indicate transition counts, not student counts. Since individuals enroll in multiple courses across disciplines within a wave, totals may exceed the sample size
First-generation status
As expected, continuing generation students (n = 24) maintained their STEM course-taking behaviors over time (Fig. 5, panel b). Nevertheless, first-generation students showed an increase from waves one to four of course transitions towards STEM disciplines. Interestingly, first-generation students enrolled in similar proportions in wave three in courses in STEM, SOCSCI, and HAS, indicating potential further interest in an interdisciplinary perspective to STEM. Similar to the case of STEM major students, we observed a pattern in which continuing generation students enrolled in subsequent DSC courses in wave two after enrolling in DSFSG.
[See PDF for image]
Fig. 5
Course-taking path after taking the online course data science for social good based on first-generation status
Note: Links represent course transitions. Numbers indicate transition counts, not student counts. Since individuals enroll in multiple courses across disciplines within a wave, totals may exceed the sample size
Race
Like the case of continuing generation college students and non-STEM majors, white students (n = 22) enrolled in courses related to DSC as a discipline after enrolling in DSFSG (Fig. 6, panel a). Mirorring the case of STEM major students, nonwhite students showed enrollment continuity over time in courses related to DSC (Fig. 6 panel b). We posit that the online project-based learning course design of DSFSG allowed learners to explore DSC further as a potential career choice and opened their perspective to other course electives they could take.
[See PDF for image]
Fig. 6
Course-taking path after taking the online course data science for social good based on race
Note: Links represent course transitions. Numbers indicate transition counts, not student counts. Since individuals enroll in multiple courses across disciplines within a wave, totals may exceed the sample size
Discussion
We examined in what ways students’ course-taking behaviors evolved after attending an online project-based learning DSC course. Specifically, students’ course enrollment data was examined using multiple Sankey diagrams to visualize students’ career choices unfolding over time. This study’s results provide insights into the course-taking behaviors of students who initially chose to enroll in an online project-based DSC course, identifying the diverse pathways that specific subgroups take after this initial exposure to DSC.
To address the first research question, we used Sankey visualizations to examine transitions among courses that students take semester after semester rather than on the traditional approach of looking at student flow over specific programs. In an era of interdisciplinarity, students’ career choices are much more convoluted than decades ago, demanding greater data granularity to discover their career choices. For example, a student majoring in STEM who enrolled in DSC elective courses might have a different career path from a STEM student whose electives centered on social sciences. Additionally, given that majors are restrictive in which classes students enroll in, centering on an elective course like DSFSG helps us see nuances in students’ agency when choosing elective courses as they advance in their preparation to enter the workforce. In this way, exploring career choices by means of course-taking behaviors could explain student’s professional identification and social contexts (Holland & Lochichotte, 2007). We acknowledge that centering our analysis on students who self-selected into an elective data science course introduces important limitations. Our findings reflect the course-taking patterns of students already interested enough in DSC to enroll in DSFSG, rather than the broader undergraduate population.
Contrary to our expectations of only observing an increasing enrollment in DSC or STEM courses, we also found an increase in enrollment toward social sciences (SOCSCI) and humanities and arts (HAS) courses in waves two and three. One could attribute an increase in interest in SOCSCI and HAS topics to sample size differences in Wave 2. However, Wave 2 was the only one in which there was no attrition. Further, provided that the ratio between STEM and Non-STEM major students is 6 to 4, one could postulate that after taking DSFSG in Wave 1, students’ interest increases toward interdisciplinary approaches to their majors, which could drive them to enroll in more SOCSCI and HAS courses. Nonetheless, unlike SOCSCI, only course transition patterns towards HAS courses survive until wave 4, which hints that HAS courses are potential hubs in higher education spaces for the exchange of ideas among fields and can promote interdisciplinary conversations. As the surge to prepare students to meet the workforce demands for data scientists continues (Coffey, 2024; Pierson, 2023), our study shows that content knowledge is also as crucial as cultivating data practices (e.g., data curation, analytical methods). We infer that for the upcoming generations of data scientists, it is not enough to have the job title of data scientists attached to their name, but also the operational fluency of the field to which they are applying DSC is necessary. Our work echoes the findings of Findley et al. (2025) qualitative study in which undergraduate students saw DSC as an expansive major that allowed them to apply the knowledge of their non-data interest, whether art or sports. From a program and curriculum development perspective, we speculate that partnerships between DSC and HAS courses could promote greater interdisciplinary and consequently increase the participation of all students in DSC.
When closely examining which courses students enrolled in wave four, we speculated about the role of online learning as a driver of interdisciplinary career paths. Specifically, the humanities and arts course in wave four was an interdisciplinary course that combined American history and environmental science and—like the DSFSG—it was taught at a distance. Therefore, while opening the doors to students from diverse backgrounds to DSC and affiliated fields (Ruthotto et al., 2021), online learning might become a vehicle to promote dialogue in apparently siloed fields. In this study, we looked at an online project-based learning course as the baseline to explore interdisciplinary course behaviors in DSC education from a descriptive perspective. Our exploratory study concurs with a previous meta-analysis indicating that project-based learning has a positive effect on achievement (Zhang & Ma, 2023). Nevertheless, our observation described course-taking patterns rather than demonstrating impacts of online format. Further, we did not look at students’ behaviors within the online course itself to understand how knowledge construction occurred when tackling real-world projects occurred. Future work could focus on the quality of collaborative knowledge construction—namely cognitive presence—by inspecting whether students go beyond stating facts to providing solutions to complex problems (e.g., Castellanos-Reyes et al., 2025.)
Contrary to our expectation, students with high achievement in DSFSG did not continue enrolling in DSC courses over time, but only those with proficient achievement did. We argue that motivation theory could explain such observation. Self-determination theory posits that individuals need a certain degree of challenge to continue engagement with an activity (Ryan & Deci, 2000). Likewise, not having enough challenging activities decreases student motivation to stay on task, resulting in an eventual activity abandonment. Building on the foundational work of Bandura (1989), students who master challenges set up higher standards for subsequent challenges. We speculate that those students who only achieved a proficient achievement continued engaging in DSC courses driven by the motivation of mastering a challenge. We suggest that a closer exploration of which other DSC courses are taken consecutively would explain the field in what ways students’ course-taking behaviors reflect—or not— a continuously more challenging path mirroring the work of Fiorini et al. (2023) contrasting the proposed and the actual curriculum. Considering that grades affect students’ GPA, which in turn might have tremendous consequences on students’ scholarships and extra-curricular opportunities (e.g., obtaining internships), it is of tremendous importance that students choose to continue engaging in DSC courses regardless of not obtaining the top grade. In terms of DSC curriculum development, we speculate that adding guidelines for advanced students could direct them to more challenging courses and increase the likelihood of further participation in DSC courses. For example, suggesting to high achieving students which courses to take after their successful completion of DSFSG.
Literature suggests that groups with limited presence in STEM education might be associated with lower participation in DSC education (Pierson, 2024), but in this exploratory study, this was not the case for all subgroups. When answering our second research question, we found that nonwhite students were continuously enrolled in DSC Education courses beyond wave two of data collection. There are two potential explanations. One is that DSC courses could serve as hooks for students with consistently less presence in STEM disciplines. The second is that continuing-generation and white students might already have a more structured career path in their minds, which might make them less prompt to explore interdisciplinary courses like those in DSC. We observed a continuous enrollment of male students in DSC courses, which was expected. However, that does not necessarily mean female students are not in DSC or related fields. On the contrary, as Pierson (2023) reported, women already make up 42–43% of bachelor’s and master’s degrees earned in statistics, closely associated with DSC, by the end of 2022. We argue that in this sample, women might already be enrolled in majors that will take them to become data scientists (e.g., statistics, computer science) and that elective courses like DSFSG might serve as catalysts for male students exploring DSC as a career. Our observation coincide by the larger proportion of female students who were enrolled in STEM courses in waves three to five. Finally, it is worth noticing that first generation college students were the only subgroup who did not continue taking DSC courses after enrolling DSFSG. Future work could examine in what ways the instructional delivery format could explain such a finding because success in online learning is related with students’ higher levels of self-regulated learning (Cheng et al., 2025). We postulate that first-generation college students have challenges acquiring self-regulation skills (Antonelli et al., 2020; Koh et al., 2022) that might not make online instruction the best fit for DSC education and consequently prevent them from moving to exploring DSC further.
Finally, we emphasize that Sankey visualizations are not intended to make causality claims but serve as valuable complementary tools for program and curriculum developers in DSC education and administrators in higher education. These visualizations generate hypotheses for future statistical verification rather than establishing causal relationships or statistical significance. From a curriculum analytics perspective (Dawson & Hubball, 2014), we argue that Sankey diagrams can reveal students’ trajectories in their path to exploring DSC. Further, Sankey diagrams allow us to observe the complexity of students’ progress in intuitive graphs (Heileman et al., 2015). Although that the results from this study are purely exploratory, we argue that Sankey diagrams offer a fast-paced exploration of the role of online education in the larger scheme of curricular decisions. The top top-down view in Sankey diagrams inform higher education institutions to continue the financial support of online learning as a format that merits attention rather than the legacy of emergency times. Further, the descriptive power of Sankey diagrams could be leveraged to inform more sophisticated statistical analyses (e.g., Fiorini et al., 2023; McFarland, 2006). We posit that there is still a need to examine whether DSC courses offered in a distance education format could increase the likelihood of students enrolling in more interdisciplinary courses from an inferential standpoint. We cannot speak to the effectiveness of online versus other delivery formats without comparative data.
Future research and limitations
Our approach of using course enrollment as a starting point, while providing valuable insights into student trajectories, limits our ability to make broader claims about DSC education effectiveness or online learning impacts. Future research should include comparison groups and students who did not initially choose DSC courses. Our subgroup analysis using Sankey diagrams resulted in hypotheses that we posit to the fields of distance learning and DSC education. We suggest two routes for future work: (1) longitudinal analysis following classic hierarchical linear modeling that follows students enrolled in multiple DSC courses over time (e.g., Castellanos-Reyes et al., 2024) and (2) taking a business analytics approach that exclusively looks at DSC courses and students’ flow among them (e.g., Fiorini et al., 2023). As Heileman et al. (2015) argued, Sankey depicted top-down insights that inform further analysis and propose hypotheses. We did not explore overlaps among subgroups, such as overlaps among students who are continuing generation but also non-white, or the case of male first generation students. We observed high similarities in the enrollment behavior of proficient achievement students and male students, which more complex visualizations could confirm. Our work exposes the need for readily available in-house dashboards that can inform higher education institutions of students’ trajectories beyond dichotomous analysis and that have affordances for intersections in groups. Further, this data has self-selection bias, given that students opting for DSC courses are more likely to pursue more DSC courses to build careers in data science and to integrate these skills into their respective fields. Specifically, those students who enrolled in DSFSG may not be representative of DSC students in general. Altogether, our aggregated approach prioritizes pattern identification over individual trajectory analysis, which is appropriate for our institutional-level research questions while protecting participant privacy.
We used course choices as potential indicators of career interest and exploration. Students’ agency to enroll in an elective DSC course suggests initial career curiosity, particularly given that DSC courses are not required in any program at our institution. However, we acknowledge several limitations in connecting course choices to broader career development. First, tracking enrollment in additional elective courses across more semesters would strengthen evidence of sustained interest rather than momentary curiosity. Second, future research could examine whether course choice patterns actually relate to occupational identity development, as conceptualized in the Career Readiness Framework (Advance CTE: State Leaders Connecting to Learning to Work, 2021). The National Center for Education Statistics’ Baccalaureate and Beyond Longitudinal Study (B&B) could enable researchers to examine whether students’ elective course patterns predict career trajectories after completing undergraduate studies. Additionally, longitudinal qualitative studies could explore how students’ course decision-making processes relate to their career development over time.
Conclusions and scholarly significance
This study has three contributions regarding pedagogy, higher education curriculum planning, and methodological implementation. At the pedagogical level, our observations indicate potential patterns of co-occurrence between enrollment in online project-based DSC courses and continued DSC course-taking patterns for all subgroups examined, except for first generation college students. Continued enrollment in DSC courses after enrolling in an online project-based learning DSC courses is prevalent among nonwhite students, students majoring in STEM disciplines, male students, and those with proficient achievement in the focal course DSFSG. This study provides descriptive insights suggesting that DSC career choices are stronger for male and proficient achievement students. In terms of higher education curriculum planning, higher education institutions that want to increase enrollment in DSC courses can see online project-based courses as a good outlet to promote interdisciplinary career paths—specifically for students who might not otherwise enroll in courses outside their majors. Not only towards DSC or STEM but also from STEM majors enrolling in SOCSCI and HAS courses. We posit that a closer look at students’ career choices based on their enrollment decisions will help us understand their exploration of DSC as a career choice.
At a methodological level, Sankey diagrams offer a fast-paced exploration of the role of online education in the larger scheme of curricular decisions. While acknowledging our findings’ exploratory and descriptive scope, we propose that using Sankey diagrams provides a dynamic and practical lens through which technology integration with pedagogy can be examined. While we have provided detailed explanations for interpreting our visualizations, we acknowledge that our novel application of Sankey diagrams to represent course transitions rather than traditional student flow may represent interpretability challenges for our readers. As visualization methods should be unambiguous, future research should include user testing to verify the correct interpretation of visualization results and validate the insights provided by this adapted methodological approach.
Acknowledgments
DCR is grateful to Dr. Sandra Liliana Camargo Salamanca and Dr. Doreen Mushi for their feedback on earlier drafts of this manuscript. We are thankful to Drs. Ray Levy and Sunghwan Byun at the Data Science and AI Academy for their support.
Author Contribution
DCR conceptualized the study, acquired, analyzed, and interpreted data, and was a major contributor in writing the manuscript. TRL contributed to data cleaning and analysis. All authors read and approved the final manuscript.
Funding
Research was funded by the National Science Foundation (# 2313644).
Data availability
Authors do not have permission to share data.
Declarations
Competing interests
The authors declare that they have no competing interests.
Abbreviations
Data science
Science, technology, engineering, and mathematics
All-campus Data science and AI Project-based Teaching and learning Course Model
National Center of Educational Statistics
Standardized mean difference
Emergency remote teaching
Data science for social good
Artificial intelligence
Educational data mining
Learning analytics
Humanities & Arts
Social Sciences
Education
Business & Economics
Health & medical Sciences
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
Advance, C. T. E. State Leaders Connecting to Learning to Work. (2021). Career Readiness Metrics Framework: A continuum of Actionable Measures of Carrer Development and Readiness.
Antonelli, J; Jones, SJ; Backscheider Burridge, A; Hawkins, J. Understanding the Self-Regulated learning characteristics of First-Generation college students. Journal of College Student Development; 2020; 61,
Bandura, A. (1989). Human Agency in Social Cognitive Theory. September.
Castellanos-Reyes, D., Camargo Salamanca, S. L., & Wiley, D. (2024). The Impact of OER’s Continuous Improvement Cycles on Students’ Performance: A Longitudinal Analysis of the RISE Framework. International Journal of Research in Open and Distributed Learning, 25(4). https://doi.org/10.19173/irrodl.v25i4.7624
Castellanos-Reyes, D., Olesova, L., & Sadaf, A. (2025). Transforming online learning research: Leveraging GPT large language models for automated content analysis of cognitive presence. The Internet and Higher Education, 65, 101001. https://doi.org/10.1016/j.iheduc.2025.101001
Cassel, B., & Topi, H. (2016). Strengthening Data Science Education Through Collaboration: Report on a Workshop on Data Science Education October 1-3, 2015. https://digital.library.villanova.edu/Record/vudl:622682
Cheng, Z; Zhang, Z; Xu, Q; Maeda, Y; Gu, P. A meta-analysis addressing the relationship between self-regulated learning strategies and academic performance in online higher education. Journal of Computing in Higher Education; 2025; 37,
Coffey, L. (2024, January 25). Data Science Major Takes Off. Inside Higher Ed. https://www.insidehighered.com/news/tech-innovation/teaching-learning/2024/01/25/data-science-major-takes-across-college-campuses
Coleman Fields, K. (2023, October 31). U.S. News Guide to College Majors. US News & World Report. //www.usnews.com/education/best-colleges/college-majors
College Majors – BigFuture | College Board. (n.d.). College Majors. Retrieved March 18, (2025). from https://bigfuture.collegeboard.org/explore-careers/majors
Davenport, T. H., & Patil, D. (2012, October). Data scientist: The sexiest job of the 21st century. Harvard Business Review. https://hbr.org/2012/10/data-scientist-the-sexiest-job-of-the-21st-century
Dawson, S., & Hubball, H. (2014). Curriculum Analytics: Application of Social Network Analysis for Improving Strategic Curriculum Decision-Making in a Research-Intensive University. Learning Inquiry, 2.
De Simon, C; Gonzalez, G; Criado Herrero, R. Network analysis: An indispensable tool for curricula design. A real case-study of the degree on mathematics at the URJC in Spain. PLOS ONE; 2021; 16,
Deming, DJ; Goldin, C; Katz, LF; Yuchtman, N. Can online learning Bend the higher education cost curve??. American Economic Review; 2015; 105,
Donoghue, T; Voytek, B; Ellis, SE. Teaching creative and practical data science at scale. Journal of Statistics and Data Science Education; 2021; 29,
Findley, K., Justice, N., Kinson, C., & Berens, F. (2025). Why swipe right?? Career interests and aspirations of incoming statistics majors. Journal of Statistics and Data Science Education, 1–12. https://doi.org/10.1080/26939169.2024.2430244
Fiorini, S; Tarchinski, N; Pearson, M; Valdivia Medinaceli, M; Matz, RL; Lucien, J; Lee, HR; Koester, B; Denaro, K; Caporale, N; Byrd, WC. Major curricula as structures for disciplinary acculturation that contribute to student minoritization. Frontiers in Education; 2023; 8, 1176876. [DOI: https://dx.doi.org/10.3389/feduc.2023.1176876]
Friedman, A., & Beasley, Z. (2024). Using textual analysis to examine student engagement in online undergraduate science education. Journal of Statistics and Data Science Education, 1–11. https://doi.org/10.1080/26939169.2024.2410796
Gasiewski, JA; Eagan, MK; Garcia, GA; Hurtado, S; Chang, MJ. From gatekeeping to engagement: A multicontextual, mixed method study of student academic engagement in introductory STEM courses. Research in Higher Education; 2012; 53,
Goodman, J., Melkers, J., & Pallais, A. (2019). Can online delivery increase access to education?? Journal of Labor Economics, 37(1).
Guo, P; Saab, N; Wu, L; Admiraal, W. The community of inquiry perspective on students’ social presence, cognitive presence, and academic performance in online project-based learning. Journal of Computer Assisted Learning; 2021; 37,
Heileman, GL; Babbitt, TH; Abdallah, CT. Visualizing student flows: Busting Myths about student movement and success. Change: the Magazine of Higher Learning; 2015; 47,
Hodges, C., Moore, S., Lockee, B., Trust, T., & Bond, A. (2020). The Difference Between Emergency Remote Teaching and Online Learning. Educause Review.
Holland, D., & Lachicotte, Jr., W. (2007). Vygotsky, Mead, and the New Sociocultural Studies of Identity. In H. Daniels, M. Cole, & J. V. Wertsch (Eds.), The Cambridge Companion to Vygotsky (1st ed., pp. 101–135). Cambridge University Press. https://doi.org/10.1017/CCOL0521831040.005
Irizarry, R. A. (2020). The role of academia in data science education. Harvard Data Science Review, 2(1). https://doi.org/10.1162/99608f92.dd363929
Koh, JHL; Herring, SC; Hew, KF. Project-based learning and student knowledge construction during asynchronous online discussion. Internet and Higher Education; 2010; 13,
Koh, J; Farruggia, SP; Back, LT; Han, C. Self-efficacy and academic success among diverse first-generation college students: The mediating role of self-regulation. Social Psychology of Education; 2022; 25,
Mashile, E. O., Fynn, A., & Matoane, M. (2023). Curriculum analytics of an open distance learning (ODL) programme: A data-driven perspective. South African Journal of Higher Education, 37(3). https://doi.org/10.20853/37-3-4835
McFarland, DA. Curricular flows: Trajectories, turning points, and assignment criteria in high school math careers. Sociology of Education; 2006; 79,
National Science Board (2024). The State of U.S. Science and Engineering 2024: Science & Engineering Indicators. https://ncses.nsf.gov/pubs/nsb20243
NC State Data Science Academy. (n.d.). ADAPT Course Model. ADAPT Course Model. Retrieved June 28 (2024). from https://datascienceacademy.ncsu.edu/courses/course-model/
Oran, A; Martin, A; Klymkowsky, M; Stubbs, R. Identifying students’ progress and mobility patterns in higher education through Open-Source visualization. 2022 IEEE Integrated STEM Education Conference (ISEC); 2022; 154, 161. [DOI: https://dx.doi.org/10.1109/ISEC54952.2022.10025315]
Pardos, ZA; Nam, AJH. A university map of course knowledge. PLOS ONE; 2020; 15,
Parker, MS; Burgess, AE; Bourne, PE. Ten simple rules for starting (and sustaining) an academic data science initiative. PLOS Computational Biology; 2021; 17,
Pierson, S. (2023, December 1). Data Analytics, Data Science Degrees See Large Increases in 2022. Amstat News. https://magazine.amstat.org/blog/2023/12/01/degreesstats2022/
Pierson, S. (2024, November 5). Data Science, Analytics Degrees See Explosive Growth | Amstat News. https://magazine.amstat.org/blog/2024/11/05/data-science-analytics-degrees-see-explosive-growth/
Rawlings, C. M., Smith, J., Moody, J., & Mcfarland, D. A. (2023). Network analysis: Integrating social network theory, method, and application with R. Cambridge University Press.
Rodriguez, I; Potvin, G; Kramer, LH. How gender and reformed introductory physics impacts student success in advanced physics courses and continuation in the physics major. Physical Review Physics Education Research; 2016; 12,
Ruthotto, I; Kreth, Q; Melkers, J. Entering or advancing in the IT labor market: The role of an online graduate degree in computer science. The Internet and Higher Education; 2021; 51, 100820. [DOI: https://dx.doi.org/10.1016/j.iheduc.2021.100820]
Ryan, R. M., & Deci, E. L. (2000). Self-Determination Theory and the Facilitation of Intrinsic Motivation, Social Development, and Well-Being. American Psychologist.
Sakamaki, K; Taguri, M; Nishiuchi, H; Akimoto, Y; Koizumi, K. Experience of distance education for project-based learning in data science. Japanese Journal of Statistics and Data Science; 2022; 5,
Tatel, CE; Lyndgaard, SF; Kanfer, R; Melkers, JE. Learning while working: Course enrollment behaviour as a Macro-Level indicator of learning management among adult learners. Journal of Learning Analytics; 2022; 9,
Wang, H., King, B., & Zhou, Y. (2019, January 25). Sankey Diagram: A Method to Visualize Student Flow and Success. ECU Student Success Conference.
Yoshia, O., Sanchez-Arias, R., & Taj, S. (2022). Application of Network Models to Assist in Data Science Curriculum Using Program and Course Learning Outcomes. Proceedings of the International Conference on Industrial Engineering and Operations Management, 2264–2273. https://doi.org/10.46254/NA07.20220490
Zhang, L; Ma, Y. A study of the impact of project-based learning on student learning effects: A meta-analysis study. Frontiers in Psychology; 2023; 14, 1202728. [DOI: https://dx.doi.org/10.3389/fpsyg.2023.1202728]
© The Author(s) 2025. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.