This project delivers an interactive text-prediction web application built using R and Shiny, satisfying the final capstone requirement for the Johns Hopkins Data Science Specialization. The application utilizes a highly compressed N-gram natural language model (built from blogs, news, and Twitter text data) and implements an efficient "Stupid Back-off" algorithm to predict the next word dynamically as a user types. The entire system is engineered for low latency and a minimal memory footprint, making it ideal for deployment on standard web environments or mobile devices. Included are the links to the live, operational interactive Shiny application hosted on shinyapps.io, alongside a compiled 5-slide markdown presentation pitching the data pipeline, algorithm structure, and performance metrics of the tool.

Exploratory data analysis and n-gram frequency visualization of the SwiftKey text dataset for the Johns Hopkins University Data Science Capstone course.