This Presentation is created to pitch an Application build to predict the Next Word while typing through Keyboard. This application built as part of the final Capstone Project for Coursera Data Science certification offered in collaboration with John Hopkins University.

Goal of Project:

Goal of project is to build Predictive Model for English Language text. Skills Required for building Predictive Model:

• Natural Language Processing.
• Text Mining.
• Data Exploration, Loading and Cleaning of Data.
• Shiny App Building Skills.

Source Data for Project is found @ https://d396qusza40orc.cloudfront.net/dsscapstone/dataset/Coursera-SwiftKey.zip

• Dataset contains the Text files recorded From US News, Blogs websites and Tweets from Twitter.
• It has Text files from English, German and French Language.
• We have built a Model only for English Language

This Data set is provided by the SwiftKey Company for developing the text prediction model.

Following steps are performed to Build an Text Prediction App:

• Text Mining : All Files are read and taken a subtastantial subset of those files. Clubbed together for Text Analysis.
• Removed Profanities i.e. Bad Words.
• Remove Whitespaces, punctuation and numbers to Clean it up.
• Tokenization : Tokenizing the data by creating 1-Gram, 2-Gram, 3-Gram, 4-Gram Datasets capturing the frequency of word combination occurrances.
• Use N-Gram | N-1 Gram, Back-Off Model technique along with N-Gram Models to Predict the Next Word.

Below is the Working Model of Next Word Predictor App

Code Repository @ GitHub : https://github.com/ravinderpratap/Capstone_coursera

• Capstone_Exploratory_NGram.R : Text Mining, Cleaning and Tokenization
• Next_Word_Predictor.R : Function for Prediction used in Shiny App by loading dataset created.
• word_predictor : Folder contains ui.R and server.R code for App.

Shiny App Link : https://rpsingh85.shinyapps.io/word_predictor/

Presentation Link : http://rpubs.com/ravinderpratap/Word_Prediction_Presentation