Data8: The Foundations of Data Science at Berkeley

As part of our week of Python, we wanted to focus on innovative pedagogical approaches that have been used to scale outreach efforts. A great example is the http://Data8.org (Foundations of Data Science) course that has been offered by the University of California/Berkeley Division of Data Sciences.

The course combines three perspectives: inferential thinking, computational thinking, and real-world relevance. Students are asked to use real data to understand relationships and patterns while teaching critical concepts and skills in computer programming and statistical inference.

The course objectives of Data 8 are wide-ranging. Upon completion of the class, students should be able to:

• Write correct small programs that manipulate and combine data sets and carry out iterative procedures.
• Extend a program with multiple functions so that it runs correctly with additional functionality.
• Calculate specified statistics of a given dataset.
• Identify the sources of randomness in an experiment.
• Formulate a null hypothesis that relates to a given question, which can be assessed using a statistical test.
• Carry out statistical analyses including computing confidence intervals and performing hypothesis tests in a variety of data settings.
• Given the result of a statistical analysis from the course, form correct conclusions about a question.
• Given a question and an analysis, explain whether the analysis addresses the question and how the analysis could change and still address the question.
• Articulate the benefits and limits of computing technology for analyzing data and answering questions.
• Correctly generate and interpret histograms, bar charts, and box plots.
• Correctly make predictions using regression and classification techniques.
• Assess the accuracy and variability of a prediction.

The creators have developed a simplified approach to teaching Python with the goal of reducing the cognitive load of standard packages such as pandas (see a parallel in R in our “Less Volume, More Creativity” entry). They write:

The course uses a module for table manipulation, charts, and maps that provides an interface appropriate for an introductory course. The Table class is similar to a DataFrame in Pandas, but explicitly does not support row indexes, hierarchical indexes, time series data, missing values, slicing, and many other advanced features that can complicate table manipulation for novices. The charting features use Matplotlib, but customize the output to match the pedagogical goals of the course. The mapping features are implemented by Folium, but aim to simplify working with tables and geojson files. While the datascience module can certainly be used outside the context of the course, it was specifically designed to support the Data 8 curriculum, while setting up students to transition to more standard tools such as Pandas.

An example of the code to bootstrap a regression coefficient can be found here. The textbook, available as a website, includes the other examples from the class.