Smartphone users frequently type phrases that contain predictable next words.

The goal of this project is to build a next-word prediction system using the SwiftKey text dataset.

The application accepts a multi-word phrase and predicts one likely next word.

The model was developed using the SwiftKey dataset containing text from:

• Blogs
• News
• Twitter

The text was cleaned and tokenized to identify word sequences.

Exploratory analysis was used to understand word frequencies and the distribution of words per line.

The application uses an n-gram based approach.

• A bigram model captures pairs of consecutive words.
• Word frequencies are used to select likely predictions.
• The unigram model provides a fallback when a suitable bigram is not available.

This provides a simple and efficient approach suitable for a real-time Shiny application.

The application provides a simple text box where the user enters a phrase.

The system processes the phrase and returns a single predicted next word.

The deployed application is available at:

https://wp2300-swasthik-kotian.shinyapps.io/SwiftKeyPrediction/