Haskell Data Analysis Cookbook

Creating n-grams from a list

An n-gram is a sequence of n items that occur adjacently. For example, in the following sequence of number [1, 2, 5, 3, 2], a possible 3-gram is [5, 3, 2].

n-grams are useful in computing probability tables to predict the next item. In this recipe, we will be creating all possible n-grams from a list of items. A Markov chain can easily be trained by using n-gram computation from this recipe.

How to do it…

1. Define the n-gram function as follows to produce all possible n-grams from a list:

   ngram :: Int -> [a] -> [[a]]
   ngram n xs 
     | n <= length xs = take n xs : ngram n (drop 1 xs)
     | otherwise = []

2. Test it out on a sample list as follows:

   main = print $ ngram 3 "hello world"

3. The printed 3-gram is as follows:

   ["hel","ell","llo","lo ...