Sarah Romanes

PhD Candidate in Statistical Machine Learning

The University of Sydney

Biography

Sarah is a PhD candidate in Statistical Machine Learning, studying under the supervison of Dr. John Ormerod and Professor Jean Yang at the School of Mathematics and Statistics within the University of Sydney. Her main research interests involve creating discriminant analysis algorithms for high dimensional data, alongside their efficient, user friendly, and fast software implementations in R. Alongside her PhD, Sarah holds a Postgraduate Teaching Fellowship at the University, and enjoys lecturing first year statistics and data science units.

Interests

Statistical Machine Learning
Bayesian Statistics
R Package Development
Data Visualisation
Statistical Modelling

Education

BSc (Advanced) (Hons I) in Mathematical Statistics, 2016

University of Sydney

Skills

R
Python
C++
JavaScript
Latex, Markdown
HTML

Packages

genDA

Multi-distributional Discriminant Analysis using Generalised Linear Latent Variable Modelling

multiDA

A discriminant analysis algorithm for high dimensional data

Recent & Upcoming Talks

multiDA and genDA: Discriminant Analysis Methods for Large Scale and Complex Datasets

Jul 12, 2019 1:30 PM

useR!2019

Slides

genDA: Using Variational Approximations to efficiently build a Generalised Discriminant Analysis Algorithm

Jun 13, 2019 1:30 PM

ACEMS 2019 Enabling Algorithms Theme Symposium

Slides

R-Ladies Sydney - Introduction to Machine Learning - Part 2

Oct 17, 2018 6:30 PM

R Ladies Sydney Meetup

Slides Code

R-Ladies Sydney - Introduction to Machine Learning - Part 1

Oct 10, 2018 6:30 PM

R Ladies Sydney Meetup

Slides Code

multiDA: Prediction of Melanoma Prognosis Class using a Multiclass Discriminant Analysis classifier with Feature Selection

Aug 1, 2018 10:30 AM

2018 Joint Statistical Meetings

R-Ladies Lightning Talk

Jun 20, 2018 6:30 PM

R Ladies Sydney Meetup

Slides

Recent Publications

SE Romanes, JT Ormerod, J YH Yang. Diagonal Discriminant Analysis with Feature Selection for High Dimensional Data. JCGS, 2019.

Preprint PDF Source Document

JT Ormerod, MS Stewart, W Yu, SE Romanes. Bayesian hypothesis tests with diffuse priors: Can we have our cake and eat it too?. In arXiv, 2017.

Teaching

I am a lecturer and tutor for the following courses at the University of Sydney:

Recent Posts

Aug 28, 2018 4 min read data-viz

My first gganimate - exploring concepts from first year linear modelling!

Have you ever had one of those moments whilst teaching where the content blows your mind? Today, whilst teaching MATH1005 at the University of Sydney, that exact thing happened to me. This weeks content was focused on teaching the students the introductions to linear modelling. A very strightforward topic, and one ususally understood well by first years. However, a twist in our teaching style left me with such an appreciation for the simpler explanations in statistics.

CONTINUE READING

Aug 8, 2018 7 min read data-viz

An argument for eating that pizza whilst on vacation - exploring the sugrrants package

Ah, back to reality! Last month, I was fortunate enough to be able to travel away from the cold Sydney winter over to Central and North America for a holiday and also to present my work on multiDA at JSM in Vancouver. Whilst at JSM, I was able to hear Earo present her work (along with Di and Rob) on making calendar plots in R, using data from pedestrian traffic in Melbourne (see more here).

CONTINUE READING

Jun 22, 2018 3 min read r-ladies

R-Ladies Sydney Launch!

The Launch! Wow, what a launch! Thank you so much Jen, Dani, Steph, and Lisa for all your work setting up this launch! I can’t wait to see what the future events will hold! I was lucky enough to be able to give a lightening talk describing my R journey below - of which I have summarised for those who could not be there! You can also find the slides here.

CONTINUE READING

Contact

sarah.romanes@sydney.edu.au
Carslaw Buildng, The University of Sydney, Australia