Categorize and predict American Twitter users' ideology with a new method

What is it about?

The division of political groups on Twitter has always been one of the hot topics in political communication, but the research in this area is often difficult to reproduce and ignores the shielding of the Internet. I will use the method of decision tree classification in machine learning to improve this research. Two decision tree classification models are constructed to research American Twitter users in this study. The first decision tree classification model has three dependent variable categories (Republican, Democrat, Independent or else), and the F1 of the model test set is 0.5. The second decision tree classification model has two dependent variable categories (Republican, Democrat), and the F1 of this model's test set is 0.856. The two decision tree classification models are a complete set of hierarchical models, and the internal comparison of two models also reveals a more distinct classification of American adult Twitter users who have a preference for "Republican" or "Democrat", and American adult Twitter users who prefer "Independent or else" are indistinguishable.

Featured Image

Photo by Brett Jordan on Unsplash

Photo by Brett Jordan on Unsplash

Why is it important?

The characteristic importance of this model also explains "the perception that Twitter is banning users' use, the impact of Twitter on American democracy, the frequency of searches for issues related to politics and society" is particularly important for Twitter's ideological categorization.