Coursera Data Science Capstone Final Project
Ozge Tugrul Sonmez
02/20/2021
Introduction and Data Preparation
The goal of this exercise is to create a product to highlight the prediction algorithm that you have built and to provide an interface that can be accessed by others.
In order to prepare data for the prediction model:
- Blogs, Twitter and News data are combined to obtain main data source.
- Character “I” is substituted by character “i” since lower() function turns the upper case letter “I” to lower case letter (“i”).
- All of the letters are converted to lowercase. All punctuation, numbers,symbols,urls are removed.
- N grams (unigrams, bigrams,trigrams and quadgrams) are obtained in order of term frequency and data frames are saved.
Next Word Prediction Model
- Input data is cleaned. Character “I” is substituded by “i”. Letters are converted to lowercase,white spaces are trimmed. Numbers, punctuation are removed.
- Quadgram, trigram and bigram data frames are obtained and terms with frequency higher than 50 are extracted.
- If the number of words entered is at least 3, then respectively quadgram, trigram, bigram and unigram frequency is used to predict next word. The most frequent 3 words are shown on the screen.
- If the number of words entered is 2, then respectively trigram, bigram and unigram frequency is used to predict next word.The most frequent 3 words are shown on the screen.
- If only one word is entered, then respectively bigram and unigram frequency is used to predict next word.The most frequent 3 words are shown on the screen.
- If nothing is entered then unigram frequency is used to predict next word.The most frequent 3 words are shown on the screen.
Shiny Application
