Text Prediction

This is a presentation of a text prediction app based on a text corpus of twitter, blogs and news articles. The project is a capstone for coursera data science specialtization and supported by SwiftKey.

Slides Over view:

• App description
• Try the app
• Advantages and Disadvantages of this approach

This app is built on a predictor function that contains:

• A split function; takes the a phras input and refere the last three, two or one word to the fetch function
• A fetch function; matches the split output with corresponding ngram and return the highest probable single word.

ngrams were built based on the frequencies of compinations of two, three or four words in the corpus

Here is a link of the app deployed on shinyapps.io

Try APP

Users enter a phrase and push the submit button in the side panel. The app returns the highest probable word in the main panel.