This is the project for the last course Developing Data Products as part of the Data Science Specialization in Coursera. This project consists of a shiny application which predicts the next word based on the given text input.

1. The algorithm is based on the Stupid Backoff Model (SBF) which selects the prediction with the highest frequency.
2. For a given input text, it will look for predictions based on trigram, bigram and unigram input.

Let’s say the input text is “try that ice”. The algorithm will first look for quadrigrams starting with “try that ice”, trigrams starting with “that ice” and bigrams starting with “ice”. If in this case, the frequencies are:

1. “try that ice and”, 2
2. “that ice cube”, 3
3. “ice cream”, 50

The algorithm will select the “ice cream” bigram and thus the next word will be “cream”.

Just enter some text in the input text box and press the “Submit” button and the predicted word should appear.

Source code: https://github.com/aneesh-arch91/data_science_capstone-word_predictor

You can download the source code and run the application locally.

Shiny Application: https://adoval92.shinyapps.io/word_predictor/