This repository contains the analysis and predictive modeling of terrorist incidents in the United States. The project leverages Python, SAS, and Tableau to develop predictive models using logistic regression and classification methods, aiming to predict attack outcomes and enhance security measures.
I undertook an extensive study of terrorism in the United States, focusing on developing predictive models to assess the outcomes of terrorist attacks. By employing logistic regression and classification techniques, we achieved a 48% accuracy rate in predicting binary outcomes of attacks. Our work also involved significant data visualization efforts using Tableau, facilitating a deeper understanding of the data through efficient preprocessing, data cleaning, and outlier detection.
Predictive Modeling: Utilized logistic regression and classification methods to predict the success rates of terrorist attacks, achieving 48% accuracy in binary outcome predictions.
Data Insights: Leveraged Tableau for enhanced data visualization, aiding in the preprocessing and cleaning of data for model accuracy.
Machine Learning: Explored various machine learning methods, including regression, classification, and clustering, to identify areas of severe attacks across states.
Model Proficiency: Demonstrated a high proficiency in data analysis, achieving approximately 84% success rate in attack predictions.
Data_cleaning.ipynb: Jupyter Notebook containing the data cleaning and preprocessing steps.
Terrorism_Cleaned_Data.csv: CSV file of the cleaned and preprocessed dataset used for modeling.
Terrorism_Dashboard.twb: Tableau workbook file with data visualizations and insights.
Project_Report.docx: Comprehensive report detailing the project's background, methodology, and findings.
USTerrorism.pptx: Presentation slides summarizing the project outcomes and managerial insights.
The project utilized the Global Terrorism Database (GTD) to analyze terrorist incidents in the United States over the last 50 years. This dataset includes detailed information on each incident, enabling a comprehensive study of trends, attack methods, and outcomes. The analysis focused on data from 1970 to 2017, with specific attention to attack types, target types, and weapon usage.
Data Preprocessing: Involved cleaning, filtering, and selecting relevant data for analysis. Significant efforts were made to address missing values and outliers.
Predictive Modeling: Employed logistic regression for predicting attack outcomes and explored the relationship between weapon types and attack severity.
Data Visualization: Utilized Tableau to create insightful visualizations, highlighting key patterns and trends in the data.
The analysis revealed critical insights into terrorism trends in the United States, highlighting the effectiveness of different attack types and the impact of weapon usage on attack outcomes. These findings provided valuable information for government agencies and policymakers in enhancing security measures and formulating strategic responses to terrorism.