Determining the Most Popular Streaming Service using Machine Learning

Abstract

Over the past years, twitter has become a popular medium for sharing views and ideas about personalities, brands, products or services. Analyzing sentiment of people to figure out the popularity of different streaming service by the twitter profiles is helpful for determining positive or negative views. This is a comparative analysis to predict or show which of the chosen streaming services is most familiar or liked by the public. To do this, different machine learning algorithms are used to computationally identify and categorize public opinions to draw a final result. The machine learning algorithms used here are Linear SVC, Naïve Bayes and Decision Tree. These help in receiving the data and predict the output within an acceptable range. The data in this case has been extracted from Twitter using Twitter API. Twitter API takes the parameters that can access many features of Twitter and also post and find tweets containing desired words. This includes data cleaning which refers to exclude the incorrect and unnecessary forms of data. This makes the way of data processing easier, faster and more compatible. On analyzing, the frequently used words are assessed. The classifying words are trained using the above mentioned algorithms. These algorithms are the supervised classifiers which are effective and efficient when the quantity of the data is huge. Using one or more algorithms helps to decide, compare and contrast the results. Once the classifiers are trained, testing is done. Testing gives the proper assessment of the data that is required for the desired results. The performance of the test set can be checked to draw a final result. Hence, comparing the results obtained for different streaming services helps to decide the most popular streaming service.