Portfolio

A collection of simple examples and links to my commercial projects demonstrating some of my data science abilities.

 
 

DATA PRODUCTS & solutions: IKIZMET.COM

As iKizmet’s Lead Data Scientist I provide ongoing end-to-end development of data products including dynamic reporting, interactive web apps and customer-facing business intelligence metrics and tools. I conceptualize and propose new products, research the primary literature and develop the necessary algorithms. I then code these into on-demand business objects for thousands of fitness studios. Examples include: financial forecasting, customer lifetime value and churn prediction, and industry peer comparison.

 

R-PACKAGE DEVELOPMENT: JSON2ENV

More and more web technologies are speaking JSON (JavaScript Object Notation) when sharing data and information. json2env is an R package I created to allow easy communication between R functions and interacting languages.

 

WEB SCRAPING & MAPPING: USED OIL OTTAWA

Governments, including municipalities, are increasingly sharing their data; however, additional interpretation and visualization are usually required to be of use to the average person.

Here, I’ve scraped a static list of waste-oil recyclers from a municipal website and created an interactive map that allows users to more easily find their nearest providers.

 

Natural Language Processing: Predictive Text

Data science is commonly used to solve a class of problems dealing with languages, known as Natural Language Processing. This includes things like translation, text-to-speech and semantic analysis.

Here I present a project I completed for the Coursera Data Science Capstone by Johns Hopkins University on Coursera (view certificate). The goal was to develop a predictive text algorithm and wrap it in a simple web app demonstration interface.

The challenge with text prediction is balancing accuracy with speed of execution.