Log Likelihood, Part 2
Natural Language Processing(Part 19)
📚Chapter 3: Sentiment Analysis (Naive Bayes)
If you want to read more articles about NLP, don’t forget to stay tuned :) click here.
We will continue from the previous Tutorial and show you how to do inference. Given your lambda dictionary the task is pretty straightforward. You have done most of the work to arrive at your log likelihood already. Now let’s wrap up. Now you can calculate the log likelihood of the tweet as the sum of the lambdas from each word in the tweet. So for word I you add 0, for am you add 0, for word, happy you had 2.2, for words because I and am you add 0 and for word learning you add 1.1. This sum is 3.3 and this value is higher than 0.
Remember how previously you saw that the tweet was positive if the product was bigger than 1 with the log of 1 equal to 0. The positive values indicate that the tweet is positive, a value less than 0 would indicate that the tweet is negative. The log likelihood for the suite is 3.3 and since 3.3 is bigger than 0, the tweet is positive. Notice that this core is based entirely on the words happy and learning both of which carry a positive sentiment. All the other words were neutral and didn’t contribute to the score. See how much influence the power words have.
Summary
Let’s do a quick recap. You used your new skills to predict the sentiment of a tweet by summing all the lambdas for each word that appeared in the tweet. This score is called the log likelihood for the log likelihood, the decision threshold is zero instead of one. Positive tweets will have a positive lock likelihood above 0, and negative tweets will have a negative lock likelihood below 0, well done. Congratulations now that you have a good understanding of how to compute the log likelihood and how to do inference, I will proceed to show you how to train a naive based model.
🚀 Want to Learn More?
📌 Follow Coursesteach for updates on machine learning, NLP, data science, and more!
💬 Are you an NLP expert with ideas to improve this blog? Drop your comments — we love contributions!
🔗 Resources and Community
👉 Course: Natural Language Processing (NLP)
📝 Notebook
💬 Need help getting into Data Science and AI?
I offer research supervision and career mentoring:
📧 Email: mushtaqmsit@gmail.com
💻 Skype: themushtaq48
🔗 Connect with Us
References
1- Natural Language Processing with Classification and Vector Spaces