The predictor takes preprocessed data searching the word most likely to be next.

Taking only the last 2 words, search the next word in a dataset of n-grams taking the most frequent word having the 2 words before

The data was taken from twitter, news and blogs. With this data, all combinations of 2 and 3 words and their frequency were searched (n-gram)

We’ll use the next sentence: “Hello pretty girl, how are”

and the app will give us the predicted word and a plot with the confidence of the word