Data Science Specialization Capstone - Let's Predict your next word

Raphael Parrado MD
08/10/2019

Objective

  • Develop a word prediction algorithm based on raw data.
  • Develop a friendly interface with an input phrase in a text box and text outputs.

Introduction

  • Texting has become increasingly common and a part of the daily life.
  • However with small phones texting large sentences become hard, slow and can hurt tendons in the hand.
  • Prediction can help resolve all these problems.

alt text

The Algorithm

  • We used the ngram model to develop the predictions for this app. Data was gathered from news, blogs and social media.
  • Data was cleaned and organized
  • Rstudio was used to develop 2-5 ngram models with the Backoff method.
  • A user friendly app was developed,

Let's Predict our next word - App and Instructions

alt text

  1. Write 1-5 words
  2. Push Predict!
  3. Enjoy!

Thank you!!!

Special thanks to the Data Science Specialization Project from Johns Hopkins and the SwiftKey Team

Shiny link:https://raphael393.shinyapps.io/LetsPredictWord/