Typing on mobile is slow and error-prone.
The goal of this app is to predict the next word a user is likely to type, based on an input phrase.
This can improve typing speed, accuracy, and user experience.

• Dataset: SwiftKey corpus (blogs, news, Twitter)
• Cleaning: lowercase, remove punctuation/numbers/profanities/stopwords
• N-gram models: 1- to 4-gram with RWeka
• Storage: Frequency tables in .Rds format

• Backoff strategy:
  • 4-gram match → if not found,
  • 3-gram → then 2-gram → fallback to unigram
• Fast and simple lookup based on string matching

Click here to open the app
1. Enter a phrase
2. Click “Predict”
3. View the next word
User-friendly interface built with Shiny

• Working prototype of a word predictor
• Core logic implemented in base R
• Potential for future enhancements:
  • smoothing
  • personalization
  • neural models / transformer-based approaches