Abstract
Abstract — Social Media Network has been an important information source, and the information propagation within the network gave an impact on politics, marketing, and entertainment industry. Our aim is to predict a tweet whether the information will be propagated further. The previous research has focused on analyzing this task with a wide range of learning methods and features, such as content and account features. Timeline features are proposed as features that can further predict information propagation and as we compared the performance with content and account features. The dataset consists of 43.229 tweets, we predict the information propagation with logistic regression, support vector machines, and random forest learning method with these features. Our result indicates that the timeline feature can be a good candidate for predicting information propagation and the random forests learning method consistently performs better. From the training result, we further calculate feature importance. Recently tweets, engagement with another user and previous liked tweets on the timeline features contributed to more popular tweets.