James McKay

The New Zealand RiverGuide was developed to provide live environmental data alongside recreational guides for activities in New Zealand freshwater. It also provides functionality for users to edit descriptions, post public notices, and log their activities. This was a personal project in collaboration with an environmental scientist with a vision to both share and collect data about freshwater recreation in New Zealand.

This project involved a React front-end application, a high-performance back end service delivering data from over 1,000 live environment data sources, and a content management system for handling user data and content.

International visitor expenditure in New Zealand has previously been measured through a survey collection. This method was not available as a result of COVID-19 and low border crossings, so we needed to use an alternative method to measure this key macroeconomic statistic.

I developed this model to estimate expenditure based on historical survey data and electronic card expenditure data. Having only card data at the aggregate level, it was important to keep the model simple while still trying to extract as much information as possible.

As New Zealand entered a lockdown to prevent the spread of COVID-19, there was a need for accessible, high frequency, and up-to-date data. With so many data sets available and coming in fast, we wanted to produce something to get data out as quickly as possible.

I developed this Shiny R application and a data transformation pipeline to easily and quickly load and update data sets from a range of different sources and in a range of different formats. The simple and clean front end was driven entirely by configuration yet was flexible enough to accommodate various types of data.

This project was handed off late in 2020 and continues to be maintained, with it now showing hundreds of different data sets on the same core architecture that I initially developed.

For my master's research, I extended work on the definition of a cosmic rest frame. Using a sample of 4,534 galaxies, I performed an optimization to find the reference frame that minimized the observed variation in the expansion of the local universe. This analysis ruled out the cosmic microwave background as the cosmological frame of rest and was published in the Monthly Notices of the Royal Astronomical Society.

One key outcome from my research as a PhD candidate was on the impact of experimental constraints on a particular class of dark matter models. These models aim to describe the nature of dark matter, and their parameters can be constrained by experimental data.

This work uses both frequentist and Bayesian statistical methods, along with sophisticated optimization algorithms, to make inferences on the viable parameter spaces for these models.