Project-based learning emphasizes that data science and AI learning experiences should approximate the workflow of data scientists. ADAPT-based courses are designed to facilitate student engagement with data science practices. The project-based learning approach also shapes how student learning is assessed throughout the course.

Overview of Project-Based Learning

Data Science Investigative Framework

To approximate the workflow of data scientists in data science and AI courses, the ADAPT model begins from a broader understanding of the practices of data scientists. Lee and colleagues (2022) offer a framework for the data investigation process, detailing the multiple elements of data scientists’ work.

Milestones

Organizing the course project into smaller, multiple milestones can gradually engage students in project-based learning experiences with timely support. These milestones also serve as a valuable reminder for instructors of what their students need to learn during specific phases of the course. The sequence of milestones mirrors the workflow of data scientists that the course is aiming to introduce.

Example: DSA 201-Introduction to R/Python for Data Science (1 Credit)

Students will develop introductory skills in R and Python needed for data science. Topics include data types, data structures, control structures, good coding practices and reproducible coding.

• Milestone 1: Data selection
  • Students select and download a dataset.
• Milestone 2: Data dive
  • Students carry out preliminary data exploration with their project data. 
• Milestone 3: Analysis plan
  • Students write an analysis plan for exploratory data analysis. 
• Milestone 4: Data analysis and storytelling 
  • Students implement their planned analyses and present findings.

See an example of instructions for a student-facing course project description with the above milestones.

Assessment – Developing a Rubric for Data Science Projects 

ADAPT assessments of student learning need to be aligned with the project-based learning approach. Common assessment tools, such as timed tests and quizzes, may not provide the necessary support for the successful progression of course projects. Incorporating rubrics for each milestone helps clear communication of expectations and provides opportunities for students to assess their own progress.

4-Step Guides to Develop Rubrics

• Step 1: Define Your Project 
  • The first step in the process is to describe the project for which you are creating the rubric, including its objectives and key milestones.

• Step 2: Identify the Criteria for each Milestone of the Project
  • List the criteria that you will consider as evidence of achieving each milestone of the project. The criteria are the features of the milestone that you will be looking for evidence. For example, finding a data set Milestone could include: a) Features of Data, b) Data Description and Source, c) Data Naming and d) Accessibility of Data.

• Step 3: Define Performance Scales
  • Describe the performance scales for each criterion. Performance scales indicate the degree to which the criteria have been met. The performance scales can range from two to five levels. Think about what evidence of a successful course project you are hoping to see in the student’s work.

Examples of Performance Levels (choose appropriate number of levels for each criterion):

• Met, Unmet
• Accomplished, Approaching, Limited Evidence
• Excellent, Good, Satisfactory, Needs Improvement, Poor
• Accelerating, Proficient, Developing, Emerging, Limited Evidence
• Innovative, Effective, Adequate, Marginal, Ineffective
• Meets All Criteria, Meets Most Criteria, Meets Some Criteria, Does not Meet Criteria, No Attempt

• Step 4: Methods of Assigning Point Values
  • Align point values with the behaviors and products you want to encourage from your students.

For further information and guidance on rubric development, visit the DELTA Rubrics best practices