Data In The Wild

An Introductory Data Science Course for the Life Sciences

Data in the Wild introduces students to data science through the R programming language and a series of ecological problems set in Antarctica. Across the course, students use data to investigate food safety, wildlife tracking, habitat relationships, and sustainable fisheries management. Along the way, they learn skills in data wrangling, visualization, statistical analysis, modeling, and interpretation.

The course is designed for students in the life sciences and assumes no prior programming experience. Lessons begin with the basics of coding in R and gradually build toward more advanced tools such as hypothesis testing, machine learning, regression, and population modeling.

The course was developed by Drs. Katy Prudic, Jeff Oliver, Keaton Wilson, and Ellen Bledsoe, and has been taught at the University of Arizona as part of the general education curriculum. Funding comes from the NSF-funded project “Building Capacity in Data Science through Biodiversity, Conservation, and General Education” (Awards 2122967 and 2122991).

Learning Objectives

By the end of the course, students will be able to:

  1. Import, organize, summarize, and work with ecological datasets in R.
  2. Create and interpret appropriate data visualizations using ggplot2.
  3. Apply statistical tools such as t-tests, ANOVA, and regression to ecological questions.
  4. Combine datasets and use basic classification approaches to support inference and prediction.
  5. Interpret population growth models, carrying capacity, and sustainable yield in an ecological management context.
  6. Communicate quantitative findings clearly using code, figures, and written interpretation.

The Narrative

Module 1: Mission Antarctica!

The team has arrived in Antarctica and needs to establish the technical foundation for all future data work. Students begin by learning how to use R and RStudio in Posit Cloud, including how to run code, create objects, work with vectors, understand data classes, subset data, and use functions. They are also introduced to the tidyverse, where they begin learning how to select, filter, summarize, group, and mutate data. By the end of the module, students are comfortable navigating their coding environment and working with simple ecological datasets.

Module 2: Good Food Gone Bad

A wave of food poisoning spreads through the Antarctic base, and the team needs to figure out why. Students investigate a potential food contamination event using descriptive statistics and data visualization to track down the culprit. Along the way, they are introduced to histograms, scatter plots, and box plots in both base R and ggplot2, and they develop an eye for what makes a visualization effective or misleading. By the end of the module, students have solved the problem and begun exploring a dataset of their own for the course’s final project.

Module 3: Follow That Seal

With the aquaponics system compromised, the team must turn to wild-caught fish to supplement their food supply, but fishing in Antarctic waters comes with risks. Leopard seals are apex predators in the area, and the radio collars used to track them have been failing unexpectedly. Students are tasked with figuring out which collar manufacturer is to blame by comparing means numerically, visually, and statistically - first using t-tests to compare two groups, then extending to ANOVA and post-hoc comparisons when more than two groups are involved. By the end of the module, students can determine whether differences between groups are statistically meaningful and apply these tools to new ecological contexts.

Module 4: Roads and Regressions

With fishing bays selected and collars in need of replacement, the team faces two new challenges: identifying the makers of a mystery batch of collars, and planning a road to the fishing sites that minimizes harm to Antarctica’s fragile ecosystem - including avoiding habitat for Antarctic hair grass and penguin nesting sites. Students learn to combine datasets using joins and binds, get an introduction to machine learning through K-Nearest Neighbors classification, and dive into linear and multiple regression to model environmental relationships. By the end of the module, students can work with data from multiple sources, fit regression models, and interpret their results.

Module 5: Sustainable Fishing

With fishing bays selected and roads planned, the team must now figure out how many fish they can sustainably harvest without depleting the population. Students are introduced to population growth models - exponential and logistic - and use them to estimate carrying capacities and maximum sustainable yield for Antarctic toothfish populations. They then learn to fit logistic growth curves to real data and compare populations statistically. Along the way, students also step outside of R to practice finding and citing scientific literature and writing a scientific abstract in preparation for the final project.

Note: Module 5 is an example of how Data in the Wild can be extended or adapted to reflect a specific disciplinary focus. This version was developed by Dr. Ellen Bledsoe and draws on their background in ecology. The course can be taught with or without this module, and instructors are encouraged to develop alternatives that reflect their own expertise or research context.