About Me

I am a Data Scientist with Fiddlehead Technology, a data science company with approximately 20 people focusing on the consumer goods industry. My expertise at the company is in time series statistical forecasting on sales data, which I perform for thousands of SKUs.

I am a former mathematics graduate student at the University of British Columbia, where I researched the mathematics underlying the quantum Hall effect (which is where the background photo for this website comes from!). I also took several graduate-level courses on machine learning, where I developed a strong passion for data science.

In terms of previous industry experience, I spent 1.5 years doing computational physics as an R&D Scientist with C-Therm Technologies, a company in Fredericton, NB, Canada which delivers thermal analysis solutions. My largest contribution was creating the regression algorithm for their "Flex Transient Plane Source" sensor. I also contributed to multiple ongoing R&D projects which resulted in some publications.

Beyond academics, I have been playing guitar for about 20 years, and I enjoy cooking, watching hockey, and playing board games.

MSc Mathematics: University of British Columbia (August 2022).

BSc Hons. Mathematics-Physics: University of New Brunswick (December 2019).

Research Interests: Machine Learning, Statistical Forecasting, Quantum Hall Effect, Quantum Lattice Systems, Iterated Function System (IFS) Fractals.

Projects

Please feel free to browse by clicking on the links to my projects, which range from personal interests to my thesis.

Master's Thesis (click here)
My research focused on the quantum Hall effect and bulk-edge correspondence in the interacting setting using quantum lattice theory.

Deep Generative Neural Networks for Geophysical Inversion (click here)
Using a β-Variational Autoencoder to achieve disentanglement and improved convexity of the geophysical inversion objective function. Investigates a 2-dimensional subsurface density model. Code available on Github here.

Variational Autoencoder Slides (click here)
A sample lecture I created for the UBC course CPSC 540 (Advanced Machine Learning). Includes a small assignment guiding you through deriving the loss function (click here).

Detecting Outlier Goal Scoring in the NHL (click here (desktop))
(Warning: works best on desktop, and can sometimes be slow to load because Heroku's free service puts the dyno to sleep after 30 minutes of inactivity, and waking it can takes a minute). A personal machine learning project I've been chipping away at using data from moneypuck. Uses LASSO, a random forest regressor, and a support vector regressor, ultimately all added to a stacking model, to attempt to learn how many goals a player "deserved to score" in any given season. In the hockey analytics community, this is known as an "expected goals model". One can then use this model to perform outlier detection; a player is considered an "overperformer" if they scored significantly more goals than the model predicted they would score, and similarly an "underperformer" is a player who scored much less than the model predicted. Often, the players which the model deemed as large "overperformers" scored significantly less goals the following season, and similarly, large "underperformers" tend to bounce back the following year. Click the link to open an application I built using my model and Plotly/Dash, deployed on Heroku. The code for the model can be found on my github here.

Vector-Quantized Naive Bayes
Implementation from scratch in Julia. Available in my GitHub repository here.

Telecom Customer Churn
A small, older project for UBC's Applied Machine Learning course predicting customer churn using Kaggle's Telecom churn dataset. Available in my Github repository here.

Movie Recommender System
A small, older movie recommendation machine learning project using collaborative filtering for UBC's Applied Machine Learning course on Kaggle's MovieLens 100k dataset. Available in my Github repository here.

Adult Census Income Classification
A small, older project for UBC's Applied Machine Learning course predicting whether income is greater than $50k using Kaggle's Adult Census Income dataset. Available in my Github repository here.

Undergraduate Thesis (click here)
My honours project on analysis and measure theory on fractals, particularly those which can be represented via iterated function systems, also known as Hutchinson operators.

Diffusion Three Ways (click here)
Solving the 2D diffusion (i.e. heat) equation via forward Euler, backward Euler, and Crank-Nicholson algorithms. MATLAB code available for the 1-dimensional and 2-dimensional cases.

2D Ising Model - Metropolis Algorithm Monte Carlo Simulation (click here)
A neat little computational physics project which runs some Ising model magnetization simluations using the Metropolis algorithm and uses Monte Carlo to estimate some physically interesting quantities. The MATLAB code can be downloaded here, or copy and pasted from the end of the article. It's been a while since I've used this code, and I don't even have access to MATLAB anymore, so I can't guarantee anything (in fact, that goes for pretty much all the MATLAB code you find on this website). Although I do remember it making some very pretty animations!

Chaos in the Driven, Damped Pendulum (click here)
Another computational physics project which uses Runge-Kutta 4 to solve the coupled differential equations governing a driven damped pendulum, which is a chaotic system. The MATLAB code can be found here.

Publications
With C-Therm Technologies: M. Emanuel, M. Bhouri, J. Furlotte, D. Groulx, J. Maassen: Temperature Fields Generated by a Circular Heat Source (CHS) in an Infinite Isotropic Medium: Treatment of Contact Resistances with Application to Thin Films, International Journal of Heat and Mass Transfer 137:677-689 (April 2019).

With C-Therm Technologies (acknowledged as a contributor): M. Emanuel, A. Emanuel: Temperature fields generated by a circular heat source (CHS): Solution of a composite solid of two different isotropic semi-infinite media, Journal of Heat Transfer HT-19-1412 (August 2019).

With C-Therm Technologies (acknowledgd as a contributor): M. Emanuel, M. Bhouri, S. Ackermann, D. Groulx, J. Maassen: Temperature fields generated by a circular heat source (CHS) in an infinite medium: Analytical derivation and comparison to finite element modeling, International Journal of Heat and Mass Transfer 126 (November 2018).