Data Science Capstone Project
Date: February 28, 2026
Prepared by: Rizky S

The Problem:
Typing on mobile devices is often slow and prone to errors (typos).

Our Solution: - A smart Shiny application that predicts the next word in real-time. - Saves user time by suggesting the most likely next word based on the context.

This algorithm is built using Natural Language Processing (NLP):

1. Data Cleaning: Removing numbers, punctuation, and profanity.
2. N-gram Model: Building frequency dictionaries for unigrams, bigrams, and trigrams.
3. Katz Back-off: Automatically falls back to simpler models if a complex match isn’t found.

The application is designed to be simple and responsive:

• Input: Type your phrase into the provided text box.
• Process: Algorithm calculates the prediction within milliseconds.
• Output: The predicted word appears instantly below the input box.

Try the app here: Shiny App Link

This application stands out because of:

• Speed: Compressed data model ensures high performance without lag.
• Accuracy: Validated through intensive testing on news datasets.
• Scalability: Can be expanded to other languages.

Thank You!