Classifying the category of any online digital material and projecting its popularity is a critical task for a variety of systems, ranging from advertising to recommendation systems to profit and revenue generation from online content. Given the large online global access to data and the ease with which online information may be generated, we are usually advised to grasp the core concept of internet popularity growth. It's critical for a variety of services, such as developing an effective caching model, viral marketing strategies, cost estimation, ad campaigns, and basic content optimization. "Trending Videos" are videos that have gained traction as a consequence of being embedded on some of the most prominent websites on the internet, as well as a high number of individuals watching the video in YouTube. The meta data for Trending videos is directly retrieved in CSV format from the crowdsourcing site Kaggle. The most popular videos from five countries have been compiled.
In this project, I use different machine learning algorithms to predict video popularity on youtube. The dataset collected from kaggle repository. The machine algorithms are trained using Logistic regression, KNN classifier, SVM, Naïve Bayes, Random Forest and decision tree after scaling the data. The evaluation measure that could help us to find the best model i.e Accuracy, sensitivity and specificity. The main aim of this project is to find the best model for predicting video popularity based on past dataset collected from kaggle repository.