Data Science Capstone Project

The goal of this application is to predict the next word for the input phrase typed by user.

• The source data consists of various corpora scraped from publicly available sources. It can be found at https://d396qusza40orc.cloudfront.net/dsscapstone/dataset/Coursera-SwiftKey.zip
• It consist of data from three different sources - blogs, news and twitter.
• Data is then cleaned and preprocessed and ngrams are formed. Milestone report depicting the same. https://rpubs.com/rish29/milestoneReport

The application is based on Katz Back-off algorithm.

• Firstly Ngrams tables are loaded.
• Once input text is obtained from user it will be compared with to text already observed in N-Gram frequency table.
• Katz back off model estimates the conditional probability of a word given its history in the n- grams. It accomplishes this estimation by “backing off” to models with smaller histories.
• All the relevant word are then extracted and present in the output.