[at]sergiosonline / Resources
For any of the below resources, please contact me in case you require any further details or clarification.
Selected Projects and Presentations
- Data Science
- Analysis of Smoking Hazard and Socioethnographic Factors
Influencing Smoking in the American Youth (Report)
- Study of uptake rate of cigarette smoking, as well as risk of consuming other tobacco products, in a representative sample of American teenagers, surveyed throughout all states.
- Bayesian Modelling of Data Correlated in Time and Space (Report)
- Harnessing the INLA framework to fit a semi-parametric temporal model to examine CO2 emissions, and a spatial correlation model to examine the usefulness of satellite data in the study of Malaria incidence in Africa.
- Fresh Over Rotten: The Value of a Critic's Pick in Hollywood (Report)
- Study of how critical perception in Hollywood movies can translate to revenue differences, based on different features
- Machine Learning
- Attention-based Implementations of RNNs for Machine Translation Task (iPython Notebook)
- Implementation of several attention-based architectures to solve a straightforward machine translation task, with a discussion of their merits and short-comings.
- Exploration of the Model-Agnostic Meta Learning (MAML) Algorithm in a Regression Setting (iPython Notebook)
- Fascinating algorithm which allows for straightforward transfer learning (task-to-task adaptation) with the only requirement being a model trainable by gradient descent.
- MNIST Handwritten Digit Classification Problem *N<P* using MCMC (Report)
- We use MCMC to overcome one of the most challenging hurdles in machine learning with reasonable success: When the number of your training cases is less than the number of parameters to be estimated.
- Probabilistic (Monte Carlo) Computation
- Bayesian Hierarchical Inference using Metropolis and Gibbs Sampling Algorithms (Report)
- Fitting a Bayesian hierarchical model with fundamental MCMC algorithms to determine production differentials, studying the chain's convergence rate and properties
- Implementations of an LCG Pseudorandom Generator, and Importance and Rejection Samplers (Report)
- Exploration of probabilistic computation and pseudo-randomness
Preferred Quick Guides
Media Presence and Features
|Featured in the 2017 American Statistics Association's Early Career Profiles (StatTrak):
|| August 1, 2017
|Featured in the University of Toronto's Career Centre as alumni of the Department of Statistical Sciences:
||August 5, 2016
|Featured in the 2016 University of Toronto's Department of Statistical Sciences' Newsletter "Converging" (P. 3, 10):
|Featured in the University of Toronto's A&S News for my team's selection as finalists the 2014 RBC NGI Challenge:
||February 14, 2014
Formatting and content of this site may be distorted when viewed on a mobile device.
*I'm hosting this site with GitHub Pages. Ask me how!*