scelarek.github.io

Featured Project

CovidCast: Predict to Protect

Time Series Forecasting • Hyperparameter Tuning • Feature Engineering • Data Wrangling • EDA

"Like a weather forecast for pandemics, COVIDCast leverages state-of-the-art machine learning and epidemiological models to deliver precise outbreak predictions."

SideBard for Google

Retrieval Augmented AI Text Generation • LLMs • Data Communication • Hackathon • Interdisciplinary Team

"Collaborated with my team, AInsight, of UX/UI, web devs, and data scientists to develop a figma prototype of a Chatbot in the Sidebar with Retrieval Augmented Generation for Google in 24hrs"

Puppularity Contest

Web Scraping • API • Regression • Classification

"Extracted "We Rate Dogs" tweets using API calls and BeautifulSoup, then evaluated comedic vs. aesthetic feature influence on engagement through regression models."

Predicting Hotel Ratings from Reviews

Feature Engineering • Classification • NLP • Hyperparameter Tuning

"Applied data preprocessing, feature engineering, NLP, and hyperparameter tuned classification models on 500,000+ hotel reviews to predict positive ratings at an accuracy of 78.4%."

West Nile Watch

Feature Engineering • Data Wrangling • EDA • Hypothesis Testing

"Deployed hypothesis tests and classification models on 18,000+ entries to pinpoint West Nile Virus hotspots and high-risk species to guide community health interventions."

DS Templates and ML Guides

Regression • Classification • Clustering • Pipelines • Grid Search

"Streamlined machine learning workflow using pipelines to grid search across the performance of various regression, classification, and clustering models on toy datasets."

Kickstarter: The Business Behind Dreams

SQL • Data Wrangling • EDA • Classification

"Crafted a thorough business report using SQL, EDA, and classification models on 150,000+ entries to set viable fundraising targets for a tabletop board game campaign on Kickstarter."

Google Engrams: Data in the Clouds

Hadoop • AWS • PySpark • EDA

"Utilized the power of Hadoop, AWS, and PySpark on 260+ million entries in Google's corpus of books to analyze the frequency of the word 'data' over the past five hundred years."