The Goal
The objective of this project is to provide a fast, accurate text prediction tool similar to mobile keyboards to improve typing efficiency.
The Data & Cleaning
- Source: HC Corp (Twitter, News, Blogs).
- Sampling: 1% of the total data for performance.
- Cleaning: Removed numbers and extra whitespace.
The Algorithm
- Uses N-gram frequency tables (Bigrams and Trigrams).
- Employs a ‘Back-off’ strategy: if a trigram isn’t found, it checks bigrams.
How to Use
- Navigate to the ShinyApps link.
- Enter a phrase in the text box.
- Click ‘Predict’ to see the results instantly.
Conclusion
- Lightweight, fast, and hosted on the cloud.
- View the app at: https://ds-capstone-azeem.shinyapps.io/Predictor_Final/