Next Word Predictor

Johns Hopkins University Data Science Capstone Project- Coursera

Varun M R

Brief Explanation of Model.

  • This prediction model works on the Markov Chain Model.

  • Text was tockenized into unigram to trigram and fitted using markovchain package.

  • The text pre-processing was done using tidy text and stringr package.

  • Machine Model estimate of each Ngram was combined using a function in which argument is processed in the order of trigram to unigram.

How it works

When you enter a word in the text input column:-

  • It searches from trigram to unigram models for suitable words and gives predictions as output.
  • If no match is found in the text corpora, most common words will be given.

Demonstration

Application link

Thank You