This curriculum for a five-workshop series introduces participants to data science without the use of any visuals, instead using the NVDA screen reader. The audience for this curriculum includes those with sensory or cognitive disabilities that make it difficult or impossible to read text displayed visually on a screen or view visual representations such as charts or graphs. The series covers the fundamentals of the Python programming language, the use of the Pandas data science library for analyzing and manipulating data, an introduction to sonification, and the use of Astronify to create and hear sonifications.
- Nonvisual Introduction to Python: An introduction to the Python programming language using the NVDA screen reader and the IPython interactive Python interpreter.
- Data Science Fundamentals in Pandas: An introduction to the Pandas data science library, with an emphasis on exploring data in the REPL using a "conversational" style and using chaining functionality in Pandas to make highly specific queries.
- Nonvisual Data Exploration and Representation: An introduction to exploring and representing specific types of data, such as categorical, numeric, and time data.
- Sonification I: Principles of Sonification: Covers the fundamental principles of sonification, including the definition of sonification, existing use cases, the most common sonification methods, parameter mapping, and audifications. This workshop will also introduce the astronify library for sonifying astronomical data developed at Space Telescope Science Institute.
- Sonification II: Creating Sonifications with Astronify: Sonify trend data prepared in previous workshops using the Astronify library developed at Space Telescope Science Institute.
Data science is typically taught in a way that focuses on visuals. Data is almost always concerned with relationships, and the most common way to represent complex relationships is through visualizations such as charts and maps. However, these relationships are not inherently visual, and can be represented in other ways, such as sound (sonification), feel (tactile graphics), or plain text (natural language or a REPL session). Further, tools such as pandas allow for flexibility in how data sets are queried and represented, and pandas readily facilitates step-by-step nonvisual data manipulation and analysis when used in accessible environments such as the IPython REPL.
While there are inherent challenges to working with data nonvisually, a variety of modern tools and the development of emerging fields such as sonification mean that data science need not be an inherently visual discipline.