Portfolio

Recent Projects

Recently, I have been working on my skills in Tableau for data visualization. Here is an example of a recent dashboard I made using Kaggle’s London Bike Share dataset and this follow along tutorial. In making this dashboard, I was able to practice creating calculated fields and dynamic parameters as well as creating a function to update the visualizations based on a selected set of data. I’m still a beginner in Tableau, but I look forward to learning more and increasing my skills!

Predicting Carbon Flux with Deep Learning

Over the summer, I was able to work full time in a research capacity. Our team was working to build various machine learning models to accurately predict carbon and latent heat flux values across 44 different sites in the NEON network. Above, I’ve included a link to the preprint of our article which is awaiting publication. In addition, here is a code sample of the deep learning model that I constructed during this research. This model was run on a SLURM cluster computer with GPU accelration using a 10-fold cross validation technique to optimize model performance while decreasing model bias. The deep learning model was built using Pytorch, and I implemented an early stopping class to prevent model over-fitting. I am currently working on writing two academic papers with a hope that both will be published by the end of the year.

Mathematical Modeling Covid Hospitalizations

In this project, I collect, clean, and analyze data relating to COVID cases in Arizona. Next, I look into using scipy packages to model the first spike in hospitalizations using the SIHR system of differential equations. Further research can be done here by implementing machine learning techniques to learn this model’s parameters.

Deep Learning for Multi-Class Classification

This project is an extension of a homework assignment from a mathematics of deep learning class. Here, I explore the use of momentum in deep learning, practice the basics of object oriented programming, and finish by building a neural network to predict classes of wine. The wine classification occurs in number 5. Here, we use Pytorch with CrossEntropyLoss to build a model that can predict between 3 different types of wines.

Various Mathematics Projects

This is a collection of projects related to my study in my Masters program in Mathematics. If you’re feeling particularly “Math-y” and want to explore some more rigorous, theoretical mathematics projects, have a look!

Page template forked from evanca