Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


40484.pdf (3.88 MB)

ETD Abstract Container

Abstract Header

A Machine Learning Web Application for Predicting Neighborhood Safety in The City of Cincinnati

Arthur, Gifty A
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1627662505704827

Abstract Details

2021, MS, University of Cincinnati, Education, Criminal Justice, and Human Services: Information Technology.
Neighborhood safety is the seed or cell state of tackling a much bigger issue in large metropolitan cities like Cincinnati. The aim of this study is to present a novel approach for raising awareness of safety of a particular location at a specific time through a web application that can be easily assessed on mobile devices such as smart phones. This development serves as a contribution to new advanced geographic information systems today that can help tackle crime prediction problems in real-time, as criminal activities continue to evolve. The work presented explores various machine learning algorithms to determine how safe a neighborhood is by recommending a `safety score’ deduced from records of property crimes within the metropolitan area. To demonstrate the feasibility of this approach, the main focus is on real property crime data from the Cincinnati Police Department for 133,246 incidents of burglary or breaking entry, theft and unauthorized use from 2010 to 2021. The dataset was extracted live from the crime management portal where it is updated daily, and re-engineered to produce victim data to extract fine-grained information such as theft in a particular suburb and the victims involved. The proposed approach falls in line with addressing a much larger socio-economic issue since many previous research efforts have tackled the crime prediction problem by focusing on historical suspect data to determine repeat offenders. This research supports the hypothesis that victim data can be mined in combination with human behavioral activity through mobile devices such as smart phones to avoid repeat victimization. The end product is a responsive web-based application using Streamlit, that combines computed geocoded address with basic demographic information to predict the likelihood of an arrest at the location at a specific time, within the greater Cincinnati area. As the target class categories in the dataset are imbalanced, SMOTE oversampling method was used to solve the classification problem. The experimental results after re-sampling the data helped Random Forest machine learning algorithm to outperform other algorithms with about 88% prediction accuracy. The main advantage in housing the machine learning project on a web server is that it provides a safety tool for users which can be easily accessed in their pocket. Machine learning algorithm offers a great predictive power for this kind of analysis and powerful web applications like Streamlit offer visual insights through the interactive interface to communicate relevant information with end-users. The application is deployed on the cloud through Streamlit’s cloud-sharing platform and made accessible on various platforms. In addition, this work provides a discussion on the implications of the findings and offers future research direction for data-driven crime analysis.
M. Murat Ozer, Ph.D. (Committee Chair)
Bilal Gonen, Ph.D. (Committee Member)
59 p.
Information Technology
crime prediction; real-time; Cincinnati; web application; Streamlit; smart phones

Recommended Citations

Citations

  • Arthur, G. A. (2021). A Machine Learning Web Application for Predicting Neighborhood Safety in The City of Cincinnati [Master's thesis, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1627662505704827

    APA Style (7th edition)

  • Arthur, Gifty. A Machine Learning Web Application for Predicting Neighborhood Safety in The City of Cincinnati. 2021. University of Cincinnati, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1627662505704827.

    MLA Style (8th edition)

  • Arthur, Gifty. "A Machine Learning Web Application for Predicting Neighborhood Safety in The City of Cincinnati." Master's thesis, University of Cincinnati, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1627662505704827

    Chicago Manual of Style (17th edition)

ucin1627662505704827
1,243
© 2021, some rights reserved.Creative Commons License A Machine Learning Web Application for Predicting Neighborhood Safety in The City of Cincinnati by Gifty A Arthur is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by University of Cincinnati and OhioLINK.