Home

Welcome to EricSayre.com. Eric C. Sayre, PhD is a statistician, data scientist, researcher, author and programmer currently living in Vancouver, BC. He began working / consulting professionally in the field of statistics in 1997, and since then has completed two graduate degrees in statistics while remaining active in the research community. He completed his PhD in the Department of Statistics and Actuarial Science at Simon Fraser University in 2009. In the past few years, Eric has become increasingly interested in the exciting field of Machine Learning / Data Science / Artificial Intelligence. In 2017 he completed Introduction to Data Science in Python by University of Michigan on Coursera, and Machine Learning by Stanford University on Coursera. Since then he has increasingly applied ML / AI approaches to actual research questions. Eric has been a statistical consultant / data scientist / collaborator for a variety of clinical and epidemiological research groups, in both health research as well as various private industry projects. On a side note, he has also delved into cryptocurrency application development, including briefly taking on a developer role on a community project.

I am experienced in a variety of programming languages, databases, statistical packages, Machine Learning / Data Science / Artificial Intelligence environments, blockchain development environments, and other software, including: SAS (expert), R, Python (see below for ML / AI specifics), Solidity, React, Rust, Parscale, C++, Octave, Microsoft SQL, VB Script, Active Server Pages (ASP), HTML, Microsoft Office (Word, Excel, PowerPoint, Access and Outlook), EndNote, to name but a few. I have specific experience with the following Machine Learning / Data Science / Artificial Intelligence environments and packages (among others): Ubuntu (preferred OS for bare metal machine learning), Jupyter Notebook (preferred), Google Colab, data science / math Python packages including Scikit-learn, TensorFlow including Keras (preferred for ANN), PyTorch (limited experience), PySpark (limited experience), NumPy, Pandas, Matplotlib, etc., with GPU acceleration where applicable via CUDA. Machine learning models / classifiers I have worked with in the field include (among others) support vector machines, decision trees, random forests, logistic regression classifiers, as well as other ensemble classifiers (besides random forests) including voting, boosting, and bagging (bootstrap aggregating). In the way of artificial neural networks (ANN), I have worked with fully connected dense neural networks (i.e., multilayer perceptron or MLP), convolutional neural networks (CNN) for image classification, recurrent neural networks (RNN) for time series / sequences and/or natural language processing.

Finally, Eric is a well-published researcher, with over 300 publications since 1997. These are a mixture of first-authorships and coauthorships on articles published in peer-reviewed medical journals, abstracts presented at scientific meetings, research reports, invited talks and his own two graduate theses. Click the tab on the left for a complete list of publications and links to the full text of his theses.

Books

(Click the tabs on the left for more information, and links to the BIGGER SAMPLE PDFs or to purchase as Kindle eBooks or paperbacks.)

The Gradual Countdown: Quit Smoking the Easy Way! The Gradual Countdown is a highly structured, methodical, easy approach to quitting (baby steps), reducing the number and portions, and how you smoke.

The Real Junk Food Diet Book v2.0 Built on psychology, metabolism and our love of junk food. Mix entire Overeating Days into your diet days, and the pounds will drop off.

Software

(Click the tab on the left for more information, an abstract, links to the complete PhD thesis and user's guide, and links to download the FULL FREE SOFTWARE.)

Variable-Weighted Ultrametric Optimization for Mixed-Type Data (VWUO-MD) In Eric's PhD research, he developed a new method of unsupervised learning (hypothesis generation) designed specifically for mixed-type data (continuous, ordinal, nominal, binary symmetric and binary asymmetric), along with data mining software to perform the analyses. Variable-Weighted Ultrametric Optimization for Mixed-Type Data (VWUO-MD) is useful in identifying new, complex relationships between variables of many different kinds, for example between a multitude of health conditions, socio-economic and geographic factors, and health services utilization patterns. VWUO-MD is a valuable tool for exploiting the increasing multitude of highly multivariate, mixed-type databases available to researchers and industry, in developing new, previously unthought-of hypotheses.

Crowdfunding

(Click the tab on the left for more information, scientific references, and a link to the full Hopeful MS Girl website.)

Crowdfunding - Hematopoietic Stem Cell Transplantation (HSCT) for Rubiana Malla (Hopeful MS Girl) We have launched a crowdfunding website for Rubiana Malla aka Hopeful MS Girl to try to get her a scientifically proven, life-saving operation. Please read her story as it will inspire great hope.

Try our Google custom search to find relevant content on EricSayre.com and the World Wide Web.

Questions? Email

Web site and all contents © Copyright Eric C. Sayre 2023, All rights reserved.