Coursera Data Science Capstone Project

Laica Noguera
August 16, 2019

Provides user the capability to input a phrase and obtain the prediction.
You can use the slider functionality to determine the number of words to be suggested.
Link to the App: https://rgnoguera.shinyapps.io/Final_CapstoneProj/

Uses Katz Back-off Approach with Good Turing Smoothing
The model itself takes a user unput text/phrase, cleans it, and depending on the length of the phrase executes a Katz back-off algorithm from the highest possible ngram, starting from the 4gram.
Katz back-off simply starts looking for a match in the highest possible n-gram, it takes all matches available there and then starts looking for extra matches that have not appeared in the higher level ngram, but appear in lower level ngrams.