As our PhD course in citizen science continues, we’ve moved from the field to our computers to explore some analysis approaches for working with citizen science data. We spent our most recent class session working through a tutorial that Ben created to introduce point process models.

So what do point process models have to say about citizen science species occurrence data? Well, as the name suggests, point process models express the process driving the distribution of points across a landscape—for instance, locations where citizen science observations have been submitted, or locations where a certain species occurs.

Several members of our group are working with this type of model to understand species distributions and characterize the various biases inherent in different citizen science datasets. For example, check out this recent paper on variation in sampling effort in a dataset on moose occurrences, led by Jorge and coauthored by other members of our citizen science group.

While there are some R packages that have streamlined this type of analysis with built-in functions, we wanted to open up the “black box”, as it were, to get a better understanding of the inner workings of these models. So Ben developed his tutorials to build models from the ground up using the RStan package in R. As someone who is new to this type of modeling, I found this bottom-up approach helpful.

We’re also finding that it’s nice to learn together as a group with diverse academic backgrounds; it’s fascinating how different the same process can sound when described from the perspectives of, say, a statistician and an applied ecologist. By working through analysis methods together, we come away with a more well-rounded picture of how the methods are viewed and applied in each of our different fields of study.

Want to learn more? Check out Ben’s tutorials here: Part 1 & Part 2