Summer 2024 - Data Analysis Techniques - Time Series Analysis of Storm-Related Data

🌪️ Analyzing the Impact of Storms on Crime & Financial Losses in Miami
Course: Data Analysis Techniques (DAT-375)
Term: Summer 2024
Tools Used: R, Time Series Analysis, tframe, tfplot

Project Overview
In this project, I conducted a time series analysis of storm-related crime data in Miami using the Storm and Climate Data Record (SCDR). The objective was to examine whether storm events from January 1, 2017, to December 1, 2019, correlated with increased crime rates and financial losses — and to provide insights that could help law enforcement make more informed decisions about resource deployment during severe weather.

Goals & Objectives:
Analyze and compare crime-related financial losses during storm vs. non-storm periods

Visualize the cumulative impact of storms on crime using time series plots

Inform predictive strategies and resource planning for the police department

Methodology:
I used R scripts and time series modeling to analyze two datasets:

crimesStormQ.csv: Crimes committed during storm events

crimesNostormQ.csv: Crimes committed outside of storm periods

Data preparation involved filtering out irrelevant variables, removing duplicates and outliers, and ensuring a clean, reliable dataset across both timeframes. I used R packages like tframe and tfplot to plot and analyze cumulative financial losses over time.

Key Findings:
Storm-related crime consistently resulted in higher cumulative financial losses compared to non-storm periods.

The visualizations showed a clear divergence between storm and non-storm losses, suggesting that storms create conditions for either more frequent, more severe, or more opportunistic crimes.

Factors such as reduced law enforcement presence, power outages, or chaos during storms may contribute to this spike.

Challenges & Limitations:
Storms were treated as a single category — further analysis could separate hurricanes, tropical storms, or flooding events for deeper insight.

External factors like economic conditions, unemployment, or emergency policies were not included.

What I Learned:
How to use time series analysis to uncover trends over time

The value of data filtering and parameter setting in quantitative research

How to interpret and communicate findings in a way that supports policy and operational decisions

Reflection:
This project expanded my understanding of how data analysis can intersect with public safety and real-world decision-making. It emphasized the role of predictive analytics in crisis planning, and showed how even seemingly unrelated data (like weather and crime) can come together to uncover powerful patterns when analyzed with precision and purpose.