Conclusions

This project has focused extensively on two main areas in school shooting incidents: (1) the way people talk about these incidents—specifically on Twitter, and (2) the factors that play into the level of media coverage these incidents receive. We explored these questions in the Naïve Bayes, Decision Trees, SVM, and Clustering sections of this website. In the ARM and Networking section, we took a different approach by analyzing the commonalities between school shooters.

In answering the first question of how people talk about school shootings on Twitter, we performed accurate predictions of tweet classifications as being either news-based or opinion-based. The implication of our ability to accurately classify these tweets is that speakers use certain words in emotional / opinion-based tweets, and certain other words in impartial / news-based tweets, that enable us to accurately identify them.

Similarly, the implication of our success in predicting the news coverage of a given school shooting is significant because of the factors that we implemented into the algorithms. We aimed to determine the level of news coverage based on the school's racial demographics, the number of fatalities, and the maturity level of the school's students. We can speculate that there is a correlation between these factors and the level of news coverage because they were effective in contributing to the predictions.

Although this project has confined to the bounds and time constraints consistent with a typical school project, we intend to improve upon it in future work. Many of the questions posed in the Introduction section of this website remain unanswered, and we intend to continue researching these questions beyond the scope of this Fall 2022 semester. Throughout our future work, we will make updates to the conclusions as needed.

Acknowledgments

We leave our thanks to the Center for Homeland Defense and Security, the Violence Project, the US Census Bureau, and the Urban Institute for providing access to data that makes projects like this possible.