The Gallery, shown in Figure 3, provides access to the aforementioned tools. Through the gallery, access is provided to the Visualizer, dataset downloads, language libraries, dataset metrics, and the Explorer.

Assignments

As part of this research, the CORGIS project was incorporated into a Computational Thinking course for non-CS majors. This course introduced basic programming skills within a data science context, which included several assignments described as follows.
The first type of assignment is exploratory analysis. These assignments were used at the beginning of the course before the students had acquired any programming knowledge. It allowed the students to familiarize themselves with the data and explore questions about trends, distributions, and relationships in the data. As an example, a student may be tasked with loading the crime dataset and identifying the downward trend of violent crime rates over time by state. Many students struggle with the different types of charts and when to use each. This assignment teaches them how to select and interpret each of the chart types.
Another type of assignment is practice problems. These practice problems incorporate basic programming skills and interaction with the datasets. They avoid the complexity of working with complete datasets by using the simper interface calls provided by CORGIS. For example a student may be asked to write a program that prints whether or not an umbrella is necessary depending on the weather in a particular city. This simply requires a function call to the Weather library, an if statement, and a print statement.
A third type of assignment is data mapping. This helps students understand the complex nature of datasets and the relationships contained therein. Students are asked to use the Explorer to create a hand-written diagram of a dataset, showing the data fields, data types, and their relationships.
Lastly, a large-scale data analysis is assigned to the students as a final project. Students are required to formulate questions about a domain of their choosing, write code to create visualizations, and then interpret the visualizations based on their knowledge of the topic. Bart et al. claim that this constitutes an authentic form of data science that gives each student the experience of analyzing real-world data that applies to each of the students respective career paths.

The Evaluation

The authors conducted an evaluation of the CORGIS project. This consists of empirical metrics of the datasets and survey results from an actual course that incorporated the CORGIS material.
The CORGIS project includes over 40 datasets. The metrics look at characteristics of the datasets such as average branch factor, number of fields, height, number of rows, size, and data types. Based on these metrics, the authors note that some datasets should have fewer fields organized into more branches. A word cloud of descriptive tags associated with each dataset is included with these metrics. The authors note that this reveals certain biases, i.e. United States is the largest word in the cloud.
The second part of the evaluation involved a survey. CORGIS was implemented in a Computational Thinking course for non-CS students who had little to no prior programming experience. There was an 80% response rate of the 50 students enrolled in the course. The survey consisted of 26 statements that the students had to rate on a 7-point likert scale. The statements were divided into two sets of five aspects. The first set represented aspects of the main course components: learning about abstraction, writing programs, real-world data, social ethics of computing, and working in small groups. The second set of statements dealt with elements of the MUSIC Model: their belief in whether they had a choice, their interest, their sense of usefulness, their sense of success, and their belief that the instructors cared. Overall, the results showed a positive response to the course goals. They indicated that working with data related to their own career paths to be more useful than writing programs. They felt empowered and successful in learning the course material. The authors include a table that shows correlation between students' intent to continue vs. components of the course with respect to motivational components. This table shows a strong correlation between students' sense of usefulness of learning to work with real-world data and their intent to continue.

Strengths

The CORGIS project  succeeds in providing an excellent set of tools and materials for an introductory computing course that will appeal to non-CS majors. The datasets and tools are a beginner-friendly way to introduce data science to students in other disciplines, while creating an interesting and authentic experience that the students find relevant.
Bart et al. do well to lay a foundation of educational theory. They analyze the Situated Learning Theory and the MUSIC Model to determine the best characteristics of learning and apply it to course material that will be relevant and engaging to non-CS majors. Understanding the reasoning behind motivated learning and applying its concepts while developing the CORGIS project will inevitably bring it closer to success.
Prior work that incorporated data science into introductory programming courses for non-CS majors included little to no evaluation of the impact of these courses on the students. Bart et al. evaluates the datasets included with CORGIS, as well as survey results from an actual course that incorporated the CORGIS material. The CORGIS project takes their research a step further by evaluating its impact on actual students attending an introductory computing course.
The large-scale data analysis assignment certainly meets the criteria of an authentic learning experience. Students choose a dataset that pertains to their respective field, write code to generate visualizations, and interpret the results. This is a task that applies the concepts of the Situated Learning Theory and the MUSIC Model quite well, and that a student should find relevant, interesting, and useful.

Criticisms

The CORGIS project addresses a legitimate need - engaging and relevant course material for non-CS majors - but the authors could have spent more time defining the problem and developing a foundation. The authors state that computing skills are required for an increasing number of disciplines, such as sciences, arts and humanities, but they don't furnish any concrete examples. Sullivan, in "A Data-Centric Introduction to Computer Science for Non-Majors", \cite{sullivan2013data} defines this much better with examples such as policymakers analyzing socioeconomic data to gain insight for decision making, retailers developing strategies based on customer purchase data, biology researchers processing large datasets to better understand cell function, and others. Hall-Holt and Sanft also define non-CS majors more thoroughly by describing them as "students who are interested in neither software development nor the frontiers of computing, yet they expect to frequently write small bits of code as part of their work in another field." \cite{Hall_Holt_2015} 
The breadth of real-world datasets provided by CORGIS is impressive and satisfies a need for relevant and interesting data. But, this also seems to present a problem of ongoing maintenance to keep the data current and relevant for the students. As time passes, the datasets will become outdated and less relevant. These datasets should be periodically updated. Also, as other disciplines develop a need for introductory computing courses and associated datasets not addressed by CORGIS, new datasets will need to be added. So, ongoing maintenance may require a fair amount of effort.
The proposed assignments for the CORGIS project jump directly from Visualizer based assignments, requiring no coding experience, to writing code in the practice problems. This step may be abrupt for many non-CS students. An intermediate step may be necessary to move learners from analyzing data via a user-friendly online tool to actually writing the code that retrieves and analyzes the code programatically. I see that the authors have created another tool outside of the CORGIS project, called BlockPy, which may serve this purpose. It is a visual web-based tool for learning Python.