Kevin Renois

Data-driven problem solver with a background in entrepreneurship and economic development. Dedicated team member who is passionate about using data to make smart decisions. I have experience in taking projects and initiatives from ideation to execution.

I live for impossible challenges.

Skills

• Data Processing
• Data Visualization
• Machine Learning
• Deep Learning
• Natural Language Processing
• Business Intelligence
• Project Management

Projects

Detecting Malicious URLs

I created a SQL database of over 1 million known benign and malicious URLs and developed a web app that can take a URL input from a user, automatically query the database to search for that URL, and tell the user whether it is safe or dangerous. If the URL is not in the database the app will automatically run it through a classification model I built using Python to predict the probability of the URL being malicious. The model was trained using Natural Language Processing and Logistic Regression and had an accuracy score of 98%.
To see view this project, click here

Mapping The Interaction of COVID-19 Cases and Wildfires Nationwide

For this project I worked collaboratively with a team of 3 to compile, analyze, and visualize data for both COVID-19 cases up to 22 October, 2020 and fire anomalies (e.g. wildfires) in 2020 up to the same date. We created a web app that displayed an interactive map showing daily reports of confirmed COVID cases layered over heat maps of fire anomalies in the continental United States. We concluded that there was a possible positive relationship between the occurrence fire anomalies and confirmed COVID cases. We speculated that this could be caused by poor protective measures during evacuations.
To see view this project, click here

Natural Language Processing Classification Model

Using Push Shift API, I created a function that pulled a total of 10,000 comments from two subreddit pages which I then combined into one dataset. Then, using Natural Language Processing and Python, I created a model that could accurately predict whether text would come from one subreddit page or the other. I designed the model to analyze language coming from both pages and train itself to learn how to distinguish whether any given language was more fit for either source. The model performed well, returning an accuracy score of about 84%.
To see view this project, click here

Hack-a-thon Project: StimuLink

I participated in a hack-a-thon with a cross-functional team of UX designers, software engineers, and data scientists (including myself). We were given an ambiguous problem and asked to find an innovative solution as a team of technologists. The problem was that many service-based small businesses have been suffering as a result of the COVID-19 pandemic and are looking for tech-based e-commerce solutions to help meet their business needs. Each team was asked to create an e-commerce solution that could help these businesses. My Team created Stimulink, a website that businesses can use to determine a range of how much money they could qualify for the if they applied for the Paycheck Protection Program Loan (PPP) form the government. My data scientist teammate and I built a predictive classification model using data from the Small Business Administration (SBA) website.
To see view this project, click here