DATA1001 Project 2 Template

Author

550664570

Client Bio

Mayors office of criminal justice (MOCJ) https://criminaljustice.cityofnewyork.us/

The Mayor’s Office of Criminal Justice (MOCJ) overlooks all the matters relating to the improvement and maintenance of the justice system. They utilise data to implement crime prevention strategies and evaluate law enforcement performance. Understanding that law enforcement alone cannot achieve public safety, MOCJ unites institutional and community stakeholders to address the structural problems that jeopardise New Yorks stability and safety. Client:

Bio:

Recommendation

MOCJ should establish focused community-based intervention programs in precincts with high complaint volumes and low arrest outcomes, especially for grand larceny and assault, in order to lower crime in persistently high-risk areas. Evidence from 2004-2024 highlights the crime hotspots where the past intervention has not succeeded witjh reducing filing complaints, underlying the need for more effective preventions strategy. Recurring crime patterns may be more successfully disrupted by tactics like youth engagement, better lighting, and neighbourhood watch programs rather than just more police enforcements.

Evidence

IDA

This report investigates crime data that is lodged in NYC using the NYPD Complaint Data. It explores the vast trends and patterns of crime reporting and how it has been altered via the integrated major prevention strategies. It examines the various offense types over a period of time to assess the potential correlarion of the program effectiveness.

Limitations and Assumptions

The NYPD complaint dataset is way too large to successfully process and display data. To ensure performance, the data was filtered from 2006-2019. Moreover, the inavlid data entries were also filtered out to maintain reporting integrity.

It was assumed that the delays over 365 days were outliers. This helped maintain consistency but may omit invalid for deeper investigation.

Investigation of the top 10 most crimes (2006-2019).

Code 
library(tidyverse)
library(lubridate)

data = read.csv("NYPD_Complaint_Data_Historic.csv")

data$CMPLNT_FR_DT = as.Date(data$CMPLNT_FR_DT, format = "%m/%d/%Y")
data$RPT_DT <- as.Date(data$RPT_DT, format = "%m/%d/%Y")
data$Year <- year(data$CMPLNT_FR_DT)
data$delay <- as.numeric(data$RPT_DT - data$CMPLNT_FR_DT)

filtered_data <- subset(data, delay >= 0 & delay <= 365 & Year >= 2006 & Year <= 2019)


crime_rate = filtered_data %>%
  count(Year) %>%
  filter(Year >= 2006 & Year <= 2019)

delay_time <- filtered_data %>%
  group_by(Year) %>%
  summarise(average_delay = mean(delay))

crime_count = filtered_data %>%
  count(OFNS_DESC) %>%
  arrange(desc(n)) %>%
  head(10)

ggplot(crime_count, aes(x=reorder(OFNS_DESC, n), y = n, fill=OFNS_DESC)) +
  geom_bar(stat = "identity") + 
  labs(title = "Crime Categories in NYC (2006-2019)", x = "Crime Type", y = "Number of Complaints") +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1), legend.position="none")

The bar chart top categorieses the crime rate in New York City from 2006 to 2019. The most frequent offences are consistently petit larceny, harassment, and assault in its various forms. While the persistence of assault and robbery indicates ongoing public safety concerns and interpersonal violence, petit larceny emphasises the prevalence of non-violent property crimes.

This pattern emphasises the necessity of focused crime prevention strategies and resource deployment, especially in sub-areas of NYC with high volumes of repeat offences.

Evolution of Crime Volume and Impact of Prevention Strategies

Code 
ggplot(crime_rate, aes( x = Year, y = n)) +
  geom_line(colour="purple", size=1) +
  geom_point(colour="black", size = 2) +
  labs(title = "Crime Complaints Trend 2006-2024", x = "Year", y = "Annual Crimes Reported")+
  scale_x_continuous(breaks = seq(2006, 2019, by = 1))+
  theme_minimal() +
  geom_vline(xintercept = c(2012, 2014, 2016), linetype = "dashed", color = "gray") +
  annotate("text", x = 2012, y = 520000, label = "Cure Violence", angle = 90, size = 3) +
  annotate("text", x = 2014, y = 520000, label = "MAP Program", angle = 90, size = 3) +
  annotate("text", x = 2016, y = 520000, label = "Cutting Edge", angle = 90, size = 3)

Code 
delay_time <- filtered_data %>%
  group_by(Year) %>%
  summarise(average_delay = mean(delay))


ggplot(delay_time, aes(x = Year, y = average_delay)) +
  geom_line(colour = "red", size = 1) +
  geom_point(colour = "black", size = 2) +
  labs(title = "Average Reporting Delay (2006 - 2019) ", x = "Year", y = "Average Delay Time (Day)") +
  scale_x_continuous(breaks = seq(2006, 2019, by = 1)) +
  theme_minimal() + 
  geom_vline(xintercept = c(2012, 2014, 2016),linetype = "dashed", color = "gray") +
  annotate("text", x = 2012, y = 8, label = "Cure Violence", angle = 90, size = 3) +
  annotate("text", x = 2014, y = 8, label = "MAP Program", angle = 90, size = 3) +
  annotate("text", x = 2016, y = 8, label = "Cutting Edge", angle = 90, size = 3)

Overall, there was a slow decline in New York City’s yearly crime complaints from 2006 to 2019. For instance, drops in crime rates coincide closely with the implementation of prevention programs like Cutting Edge (Operations Cutting Edge, 2016), the (Bridge Initiative Team, 2020), and Cure Violence (Slutkin, 2012), indicating that these interventions may have had a quantifiable effect on reducing crime.

In a similar vein, between 2006 and 2016, the average reporting delay per incident stayed largely constant, ranging from 4.5 to 5.5 days. 2018 sees a noticeable spike that might requires more research and data gathering, but 2019 sees a sharp decline in the delay. With the MAP and Cutting Edge programs displaying not much of an impact on the reporting pattern, this pattern suggests overall progress in promoting prompt crime reporting but also indicates that some major prevention strategies did not promptly reduce the report delay.

Code 
# Create summary of delay times by offense
delay_brief <- filtered_data %>%
  group_by(OFNS_DESC) %>%
  summarise(
    median_delay = median(delay, na.rm = TRUE),
    mean_delay = mean(delay, na.rm = TRUE),
    Cases = n()
  ) %>%
  arrange(desc(mean_delay))

# Select top 5 offenses with highest mean delay
top_offenses <- delay_brief %>%
  slice_max(mean_delay, n = 5) %>%
  pull(OFNS_DESC)

# Filter data to only those top 5 offenses
delay_data <- filtered_data %>%
  filter(OFNS_DESC %in% top_offenses)

# Boxplot of delays by offense
ggplot(delay_data, aes(x = reorder(OFNS_DESC, delay, FUN = median), y = delay)) +
  geom_boxplot(outlier.shape = NA, fill = "lightpink") +
  coord_flip() +
  labs(title = "Delay Time for Offense Types", x = "Offense Type", y = "Reporting Delay (days)") +
  theme_minimal()

More intense analysis was ammended about major categories of crime which had the highest delay time to further support the MOCJ’s understanding of which offenses require targeted improvements in the reporting structure. Criminal Mischief & Related Offences had the longest average reporting delay of more than 59 days, followed by Theft of Services, as indicated in the above table and boxplot.

This breakdown demonstrates that some non-violent crimes experience significant delays, most likely as a result of victims’ lack of urgency or underreporting. These results point to the need for better public awareness and community engagement initiatives to address delayed reporting, particularly for infractions that are frequently disregarded but nevertheless have an impact on local resources and neighbourhood safety.

Ethics Statement

This report maintains integrity by providing a transparent dataset visualistaion of NYPD Crime trends. In adherence to the ISI’s principle of Pursuing Objectivity (International Statistical Institute, 2023), this report employs suitable statistical techniques, admitting data limitation, and making sure interpretations are supported by evidence to assist MOCJ with making well-informed decisions.

AI usage statement

Includes an AI usage statement covering:

  1. Name and version of the AI tool.
  2. The date/s of use of the AI tool.
  3. A link to the entire prompt session used to inform report. (In chat gpt to navigate to the conversation you want to share, click the “Share” option, and then copy the generated link.)
  4. A short summary explaining how it was used and identifies which section(s) in the research paper containing the AI tool’s outcome.

If no AI is used in the generation of the report this must be made explicit.

When asked ChatGPT, “for NYPD crime datat set, what are some major inquiry questions” - https://chatgpt.com/c/68272a17-c100-8001-8951-23001b7aed33 When asked ChatGPT “what is the program line for adding annotation on a line graph on R programming” - https://chatgpt.com/c/68272ad6-746c-8001-ad24-109a3c6a1368 When asked ChatGPT, “how to convert data into filter into year and categories as year only” - https://chatgpt.com/c/68272ad6-746c-8001-ad24-109a3c6a1368

Acknowledgements

International Statistical Institute. (2023). Declaration on professional ethics. https://isi-web.org/declaration-professional-ethics isi-web.org

Cure Violence Global. (2016). Cure Violence Global. https://cvg.org/

Wikipedia contributors. (2012). Operation Cutting Edge. Wikipedia. https://en.wikipedia.org/wiki/Operation_Cutting_Edge

Bridge Initiative. (2014). Factsheet: The NYPD Muslim surveillance and mapping program. Georgetown University. https://bridge.georgetown.edu/research/factsheet-the-nypd-muslim-surveillance-and-mapping-program/