The app works in the following manner: - Takes an input from the user, tokenises and stems it in the same manner used earlier, and then retrieves the last two words. - The last two words are used to search for matching trigrams, and if any found, the most probable ones are returned (depending upon user's desire) - If no predictions are found in the trigram model, or not enough predictions (depending upon user's request) are found, then we search for matches in the bigram model in the same manner. - At last, the unigram model is used, which does nothing but returns the required number of words randomly from the 50 most probably words in the unigram model. The randomness was introduced to create some variety in the predictions of unigram model.

Building Data Products Week3 Assignment: Time series using plotly

Building Data Products Week2 Assignment: Top 5 Places in Bangalore