Pates, Nicholas J

Introduction:

My name is Nicholas J. Pates. I am new Assistant Professor in the Department of Agricultural Economics

here at the University of Kentucky. Much of my research work involves modeling how land use and land

values vary across the contiguous United States at the field level using a variety of discrete choice models,

quasi-experimental methods mostly involving variants of difference-in-differences (DID) and unsupervised

learning methods. These projects involve modeling and replicating for statistical significance across millions

of individual fields within the US and require an HPC for handling and modeling purposes. I plan on using

the cluster to continue my research and perform similar research to construct heterogeneous farmland value

predictions, determine the spatial extents over which crop choices are made, and better understand the

observational temporal dependence in yields and crop choices.

I will likely use the cluster for the following projects using the following methodology. As this research is

ongoing, methods and applications may be altered, abandoned or expanded upon:

Defining Observations: A Study of Spatial Crop Choice Decision Boundaries

Methods:

Canny and Sobel edge detectors

Logit/Probit, Multinomial Logit, Mixed Logit

Wild score bootstrapping

Personnel:

Dr. Nicholas J. Pates (PI)

Estimating Heterogeneous Corn Supply Response to Price

Methods:

Logit/Probit, Multinomial Logit, Mixed Logit

CART, Bagging, Boosting/Gradient boosting

Wild score bootstrapping

Personnel:

Dr. Nicholas J. Pates (PI)

Estimating the Yield Impact and Memory of Crop Rotations Across the Contiguous United States 

Methods:

OLS/Linear Regression Designs

Double-Selection LASSO

Local Principal Components Analysis

Personnel:

Dr. Nicholas J. Pates (PI)

An Analysis in Agricultural Land Valuation

Methods:

Logit/Probit, Multinomial Logit

Multi-level modeling

Bootstrapping

CART, Bagging, Boosting/Gradient boosting

Propagation-Separation approach

Personnel:

Dr. Nicholas J. Pates (PI)

Dr. Tyler B. Mark (Co-PI)

Software:

R

Python

Geospatial Data Abstraction Library (GDAL)

Collaborators:

Dr. Nathan P. Hendricks (Kansas State University) (Co-PI)

Grants:

Publications: