Statistical Modeling

Our team has experience in both frequentist and Bayesian statistics as well as machine learning and can help determine the appropriate analyses for your data and write robust, reproducible code to run those analyses.

Relevant projects

A conceptual graphic of the proposed data analysis pipeline. Each step is represented by a stack of maps of the continental US. The top reads: "For each year, 1981-2023; geographic extent of all maps: NE US". Then, from left to right: "Inputs: Daily mean temperature (source: PRISM)", "Intermediat products: DOY maps for GDD thresholds", "Output: Mean DOY for each threshold, 1991-2020". Pointing downward from the intermediate products is an arrow labeled "Calculation of trends and variation for each thres

Estimating Trends in Phenology

In collaboration with the National Phenology Network, we are creating an analytical pipeline to estimate spatial trends in the time to reach different thresholds of growing degree days in the Northeastern USA.

A map of Arizona where the standard deviation in AGB is represented by color. There is high standard deviation in areas that roughly correspond to mountain ranges and low standard deviation in lower elevations.

Estimating Uncertainty in Carbon Stores

We are working with David More and Yang Li of the School of Natural Resources & the Environment to build an analytical pipeline characterizing uncertainty in estimates of above-ground biomass, to help stakeholders make informed decisions about which data products to use for their needs.

biomass level1

Statistical Analysis of Rangeland Restoration Experiments

We designed hierarchical Bayesian analysis to facilitate a paper on postfire effects of rangeland restoration treatments.

homepage of rangedocs

RangeDocs

Automated categorization of rangeland science publications using natural language processing to help rangeland managers find the information they want.

predictive rates

Trends in US Organic Agriculture

We developed a multivariate hierarchical Bayesian regression model that estimates trends by state, handles missing data, and predicts future growth of organic farming in the US from USDA survey data.

SENTINEL Phytosensors

As part of the SENTINEL: SENsing Threats In Natural Environments project, our group conducted biophysical and ecological modeling to predict the functioning, ecological impact, and physiology of plants that have been genetically modified to act as sensors to detect chemical and biological threats.

Team members

Renata Diaz

Jessica Guo

Data Scientist

jessicaguo@arizona.edu

Eric Scott

Scientific Programmer and Educator

ericrscott@arizona.edu

Pronouns:

he, him, his