Jason Richardson, PhD
Data Scientist
I am a data scientist passionate about innovation, automation, and uncovering hidden patterns in data to make business impacts. I enjoy working closely with cross-functional business leaders to identify and realize areas where analytics, ML, and AI can save costs and time and improve processes.
I earned a PhD in Biology and Evolutionary Biology from the University of Florida in 2018. During graduate school I was exposed to a great deal of statistics and programming and fell in love with it. I have been working as a data scientist since then, developing hypotheses, conducting experiments, and analyzing results.
I later worked with Coke Florida, fine-tuning my skills in a business setting. You can learn more about my work on the Projects page. Currently, I am working at Grow Financial FCU performing advanced business analysis using various techniques.
Projects
Coke Florida Customer Churn Model
When I was at Coke Florida, we had an issue where we were seeing an uptick in the number of customers in the small independently owned convenience store channel no longer purchasing from us. I was tasked with figuring out why this may be. After some exploratory data analysis to get to the root of the issue, I decided to take it a step further and build out a model that could help to predict over the next 3 months would a customer continue to purchase from us or not. This model looked not only at sales metrics but at customer staisfaction metrics as well. For example, if we had equipment in the store how often did it breakdown. If it did breakdown how fast were we to get out there and repair it? I used XGBoost to build a binary classification model. During training, I achieved an AUC of 0.89. This model was deployed in Azure ML as a scheduled batch inference pipeline that generated a weekly churn prediction list for customers in this channel. The output was automatically delivered to the field sales team via email, enabling targeted outreach to at-risk customers. Over time we saw a downturn in the number of churning customers.
Selectapref R Package
In ecology, we have a certain kind of experiment called cafeteria-style feeding experiments. These can be used to assess dietary preference in animals. There are a number of ways to analyze these experiments, but prior to Selectapref there was no central repository of these analyses, they had to be done either from scratch in a coding language or "by hand". Selectapref was my first foray in the world of open-source programming and sharing packages/libraries with others. This package has been used by others for a wide variety of species small and large including:
and MORE
The Effects of Gopher Tortoise (Gopherus polyphemus) Herbivory on Plant Community Composition and Seed Germination
This is my dissertation research and where it all started for me. I had always bee interested in reptiles and wanted to work on invasive species ecology. We have no shortage of invasive reptiles in Florida. After doing some initial research my study system shifted to our native Gopher tortoise and the impacts they have on plant communities (what species do we see and in what numbers) and the impacts of gut passage on seed germinability (seed germination rates and percentages). This is, to my knowledge, the first study of how reptilian foraging impacts plant communities. You can check out the publications on this work HERE and HERE .
You can also see my highly outdated code used to analyze these experiments HERE and HERE. These days when I use R, I use the tidyverse, which has been game changing.