Contents
Once all the conditional probabilities of each n-gram are calculated from the training text, we will assign them to every word in an evaluation text. Furthermore, the probability of the entire evaluation text is nothing but the products of all n-gram probabilities:
As a result, we can again use the average log-likelihood as the evaluation metric for the n-gram model. The better our n-gram model is, the probability that it assigns to each word in the evaluation text will be higher on average.
Please check Gaussian Sample and Bayesian Statistics.
We are looking for Data Scientist and Engineer.
Please check our Career Page.
Please check about experiences for Data Science Project
Please also visit Vietnam AI Lab
