Intoduction

This is the final capstone project of Data Science Specialization Course offered by John Hopkins University.
This project is inspired by Swiftkey smart keyboard which is available on both ios and android. The objective of this project is to build a next word prediction model with R.

Algorithm Used

I used the data provided in the coursera which contains the files:news,blogs and twitter posts. In this predictive model, it takes the user input text and try to predict by dividing into “n-grams”. For better accuracy, i used Jelinek-Mercer smoothing in this algorithm by combining the unigrams,bigrams and trigrams probabilities.

ShinyApp

-In this web app, you will see the text box where you can type. -After you type, there will be 3 predictions boxes under the text box. -You can either click one of the three buttons or if your desired word is not there, you can continue typing another word.
-If you click the button, it will automatically add the word to the text box you were typing and predict another 3 words.

Advantages and source codes

-The prediction is almost instantly after you type and does not require too much ram. -You can add the predicted word directly by clicking the buttons.

Source codes- “https://github.com/NgaMite/Swiftkey-Capstone-Project”
ShinyApp - “https://bomite.shinyapps.io/Predictive_Keyboard/?_ga=2.269073719.753087871.1601373665-1926060627.1600951877”