AMS assistant professors Lee and Guhaniyogi chosen as Hellman Fellows

Assistant Professor Juhee Lee
Assistant Professor Juhee Lee
Assistant Professor Rajarshi Guhaniyogi
Assistant Professor Rajarshi Guhaniyogi
Wednesday, April 20, 2016
Shannon Bowman

Applied Mathematics & Statistics Assistant Professors Juhee Lee and Rajarshi Guhaniyogi have been selected as 2016 Hellman Fellows. The Hellman Fellows Fund provides funding to assistant professors for their promising research. Lee and Guhaniyogi will put this money toward their respective research into alternative statistical prediction methods.

Anyone with basic knowledge of probability theory gleaned from high school statistics is familiar with the normal or Gaussian distribution. It is often used in data analyses to represent or approximate a distribution of random variables encountered in the real world. Lee, however, feels that this approach is not always proper. Instead, she has developed and proposed a different way of analyzing data by taking a  Bayesian nonparametric approach. This approach does not make an assumption on the form of distributions  on parameters  (such as  a bell shape curve in a normal distribution) so it provides a way of defining a very flexible model. Lee’s model provides more precise inference on the underlying science.

Lee has been collaborating with UCSC Ocean Sciences assistant professor Marilou Sison-Mangus to study oceanic bloom events. Bloom events release toxins into the ocean, affecting sea life and the creatures that eat them, including humans. Understanding how the bacterial communities change related to these toxic events and some environmental variables such as water temperature is of great interest to ocean scientists. . Lee will analyze the data collected by Sison-Mangus with the Bayesian nonparametric model she developed. Sison-Mangus has also provided data on gut bacteria from daphnia, which are small planktonic crustaceans that live in the ocean. Lee’s analysis will examine if the composition of bacteria community  differs between daphnia phenotypes and/or genotypes. While the data that Lee has been using is from ocean sciences, she believes that, with modifications, her model can have applications in many different disciplines.  

Lee plans to use the money provided by the Hellman Fellows Fund to support a graduate student researcher. She will also use whatever money is left over to travel to conferences in order to share her research with both the statistical community as well as ocean science researchers.

Guhaniyogi was also awarded funding from the Hellman Fellows Fund for his novel statistical research. Guhaniyogi’s work is concerned with massive spatial data, like those from geographical information systems, and how to effectively analyze and apply it to real world situations. Currently, existing models are unable to analyze big data because computers aren’t powerful enough to invert these large matrices. Building a computer that would be powerful enough to analyze data this big would be very expensive.

Guhaniyogi has a different approach. Instead of relying on a super expensive supercomputer, he plans to divide massive data into smaller subsets which he can parallelly analyze on several different computers. He can then perform an aggregate meta-analysis of these subsets and form estimates and predictions using spatial meta-kriging. Because Guhaniyogi’s method uses existing technology, this method is a much more inexpensive way to tackle the problem of big data analysis.

The data that Guhaniyogi will analyze comes from two sources: NASA and the High Plains Aquifer. The NASA data is concerned with water vapor and cloud temperature relationships and what that means for weather in different locations. The HPA is an irrigation water system that has been steadily losing water for a period of time. Guhaniyogi hopes that his analysis can begin to uncover how and why water is being lost.

Like Lee, Guhaniyogi plans to hire a graduate student researcher with the money he has been awarded from the Hellman Fellowship. He also plans to travel to conferences in order to share his research and hone his ideas. Finally, Guhaniyogi would like to make a software package so that people can more easily use his statistical methods with their own data.

Both Lee and Guhaniyogi are excited to continue their research with the funds provided by the Hellman Fellowship. Their novel solutions to complex statistical challenges sets them apart and shows how the Jack Baskin School of Engineering faculty make strides in their research fields every day.