There is a general perception that fancy multimedia articles with nice design look more special and therefore attract more engagement from audiences. There is also a view that images and video increase engagement. But do audiences are actually spend more time with a story just because it’s a special product? How much does the template matter? What about other variables, like images and video? Here we use predictive models and Shapley Additive Explanations (SHAP) to try to understand what story elements drive engagement time. About 1600 articles from the past five years were analysed. Engagement minutes (page views average time on page) is used because it is the standard metric of audience engagement.
The data is not included here as it is proprietary. The code lists the features of the data -- I'd suggest replicating this with synthetic data.
You're free to use this code however the data is proprietary.
- Email: pauldonoughue@gmail.com
- GitHub: https://github.com/DonoughuePWD