Tracing the Public Safety Implications of Privately Manufactured Firearms

Overview

This comprehensive study investigates the complex relationship between civilian firearm ownership, firearm manufacturing characteristics, and public safety indicators. Initial regression analyses of 2017 national data revealed a statistically insignificant correlation between civilian firearm density and crime indices (correlation coefficient: ±0.11), with machine learning predictive models demonstrating limited explanatory power (adjusted R² range: -0.074 to 0.24). The research critically examined privately manufactured firearms (PMFs) and miscellaneous firearm components in relation to mass and school shooting incidents. Notably, statistically significant correlations emerged between miscellaneous firearm manufacturing and shooting frequencies: school shootings (correlation coefficient: 0.82, p < 0.001) and mass shootings (correlation coefficient: 0.76, p < 0.001). Furthermore, multivariate analyses confirmed that PMFs recovered from crime scenes significantly correlate with increased shooting incident rates (p < 0.005). To enhance firearm identification capabilities, we deployed advanced machine learning techniques, including Convolutional Neural Network (CNN) and pre-trained Very Deep Convolutional Networks (VGG-16) for image classification. While model performance varied, the CNN model achieved precision ranges of 0.49-0.54 and recall of 0.31-0.74 across different classification categories. The VGG-16 model demonstrated comparable performance, with precision between 0.47-0.50 and recall ranging from 0.22-0.75. Supplementary Google Trend analysis provided additional insights into civilian firearm customization patterns across diverse geographical contexts. These findings underscore the complex interplay between firearm manufacturing, ownership dynamics, and public safety implications.

Competition history

• AJAS 2025 Category not listed

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• AJAS None projects
• Every AJAS 2025 project