Typing on mobile devices is slow and error-prone.
Can we help users by predicting the next word as they type?

We built an n-gram-based prediction model trained on real text (blogs, news, Twitter).
The app predicts the next word based on the last 2 or 3 words of a phrase.

• Text is cleaned and tokenized
• If 3+ words are typed → use trigrams
• If 2 words → use bigrams
• If no match → return “No prediction found”

• Built with R and Shiny
• User enters a phrase
• The app returns the most likely next word
• Deployed at shinyapps.io

• Improve model using smoothing or backoff
• Expand dataset for more realistic predictions
• Explore deep learning approaches

Thank you!