Uncovering the Relationship between Online News Characteristics and Popularity
Given online news’ tremendous popularity and its potential societal impacts, this study seeks to understand which characteristics within the text of a news article corresponds to a higher number of shares. Using a data set obtained from the UC Irvine Machine Learning Repository that contains information on 39,644 articles published on Mashable, we utilized 58 predictive attributes to determine which had the greatest impact on the variable of interest. AIC and BIC stepwise regression were used to determine the best multiple linear regression model, and a regression tree was created to provide extra information about which variables are most useful. This final model suggests that day of the week, category, subjectivity, and amount of positive words are key characteristics of online news articles.