Grade Level at Time of Presentation
Senior
Major
Finance, Economics, and Business Data Analytics
Institution
Western Kentucky University
KY House District #
CD2
KY Senate District #
CD2
Faculty Advisor/ Mentor
Lily Popova Zhuhadar
Department
Business Data Analytics
Abstract
This paper investigates the effects that 12 different terrorist attacks had on consumer sentiment, using data scraped from Twitter to determine a broad based emotional stance. This was inspired by my previous research that worked to determine the impact that terrorism had on the stock market. The goal of this research is to determine a more qualitative impact of terrorist attacks. I utilized Rapid Miner for data processing, and the Global Terrorism Database as my data source. The project began by examining the rise and fall of keywords and hashtags through sentiment analysis to measure reaction over time. The text was then clustered into keyword families to determine the key conversation topics. Finally, the text was analyzed to determine the evolution of polarity (the transition from negative to positive sentiment over time). The results from this analysis concluded that most of the messages were overwhelmingly negative, but shifted more towards a neutral stance as time passed after the attack. The number of tweets per day actually increased after the attack, with most tweets occurring in the early morning hours, and declining around 5 pm. An analysis of the clusters reveals a relationship between different keywords, with political keywords (i.e. Trump, Violence, Terror) often forming the strongest cluster. Consumer sentiment seemed to neutralize over time, suggesting the possibility of desensitization or a numbing effect.
The Impact of Terrorism on Consumer Sentiment: Evidence from Twitter Data
This paper investigates the effects that 12 different terrorist attacks had on consumer sentiment, using data scraped from Twitter to determine a broad based emotional stance. This was inspired by my previous research that worked to determine the impact that terrorism had on the stock market. The goal of this research is to determine a more qualitative impact of terrorist attacks. I utilized Rapid Miner for data processing, and the Global Terrorism Database as my data source. The project began by examining the rise and fall of keywords and hashtags through sentiment analysis to measure reaction over time. The text was then clustered into keyword families to determine the key conversation topics. Finally, the text was analyzed to determine the evolution of polarity (the transition from negative to positive sentiment over time). The results from this analysis concluded that most of the messages were overwhelmingly negative, but shifted more towards a neutral stance as time passed after the attack. The number of tweets per day actually increased after the attack, with most tweets occurring in the early morning hours, and declining around 5 pm. An analysis of the clusters reveals a relationship between different keywords, with political keywords (i.e. Trump, Violence, Terror) often forming the strongest cluster. Consumer sentiment seemed to neutralize over time, suggesting the possibility of desensitization or a numbing effect.