Curbing Curb Crime: Addressing NYC juvenile crime rates

Client Bio

NYPD’s Bureau of Youth Strategies Division

Dedicated to providing support and programs for young people at risk of being influenced by crime. They seek to improve relationships between police and the community while offering a place for young people to mature unaffected by gangs or crime in their respective neighborhood.

https://www.nyc.gov/site/nypd/bureaus/administrative/youth-services.page

Recommendation

After careful consideration of the data presented, it is recommended that the NYPD’s Bureau of Youth Strategies Division increase its ‘after school’ youth programs to combat a surge of juvenile crime affecting the 40th, 44st, 73rd and 75th precincts after school hours. Furthermore, these programs should provide a method of transportation across New York City, to decrease the likelihood of crime happening on the unsupervised and dangerous streets of the city.

Evidence

Areas in NYC most affected by youth crime

CRIMERECENT <- read.csv("CRIMERECENT.csv")
library(plotly)
suppressPackageStartupMessages(library(dplyr))

suppressPackageStartupMessages({
  library(knitr)
  library(kableExtra)
})

library(dplyr)
library(ggplot2)
library(plotly)

juvenile_precincts <- CRIMERECENT %>%
  filter(VIC_AGE_GROUP == "<18") %>%
  filter(!is.na(ADDR_PCT_CD)) %>%
  group_by(ADDR_PCT_CD) %>%
  summarise(juvenile_complaints = n())
threshold <- quantile(juvenile_precincts$juvenile_complaints, 0.95)
juvenile_precincts <- juvenile_precincts %>%
  mutate(danger_zone = factor(
    juvenile_complaints >= threshold,
    levels = c(FALSE, TRUE),
    labels = c("Other Precincts", "Danger Zone (Top 5%)")
  ))

p <- ggplot(juvenile_precincts, aes(x = ADDR_PCT_CD, y = juvenile_complaints,
                                    text = paste("Precinct:", ADDR_PCT_CD,
                                                 "<br>Complaints:", juvenile_complaints))) +
  geom_point(aes(color = danger_zone), shape = 4, size = 2, stroke = .5) +
  geom_hline(yintercept = threshold, linetype = "dashed", color = "red") +
  scale_color_manual(values = c("blue", "red")) +
  scale_x_continuous(
    breaks = seq(0, max(juvenile_precincts$ADDR_PCT_CD, na.rm = TRUE), by = 5),
    limits = c(0, max(juvenile_precincts$ADDR_PCT_CD, na.rm = TRUE) + 5),
    expand = c(0, 0)
  ) +
  labs(
    title = "Juvenile Crime Complaints by NYPD Precinct with 'Danger Zones'",
    x = "Precinct Number",
    y = "Number of Juvenile Crime Complaints",
    color = NULL
  ) +
  theme_minimal(base_size = 11) +
  theme(
    legend.position = "top",
    legend.title = element_blank(),
    legend.text = element_text(size = 12, face = "bold"),
    plot.title = element_text(hjust = .05, size = 12, face = "bold"),
    axis.text.x = element_text(angle = 45, hjust = 1)
  )

ggplotly(p, tooltip = "text")

This scatter plot shows that there are four NYPD precincts in the top 5% of Juvenile Crime Complaints, these being precincts 40, 44, 73 & 75. They make up ~11.86% of all juvenile crime committed in the data set. Precincts 40 and 44 are geographically adjacent (The Bronx) to each other, as are precincts 73 and 75 (North East Brookyln). Indicating that these precincts are the areas in New York City most affected most by juvenile crime.

library(kableExtra)
library(knitr)

poverty_da <- read.csv("NEWP.csv")

kable(poverty_da[1:8, ], format = "html") %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed", "responsive"), full_width = FALSE) %>%
  row_spec(2, background = "#FFF3CD") %>%
  row_spec(8, background = "#FFF3CD")

Borough	Neighborhood	X5.Year.Avg.Poverty.Rate	Margin.of.Error
Bronx	Morris Heights, Fordham South, and Mount Hope	35.4	+/-2.2
Bronx	Concourse, Highbridge, and Mount Eden	33.6	+/-2.2
Bronx	Hunts Point, Longwood, and Melrose	31.5	+/-2.2
Bronx	Belmont, Crotona Park East, and East Tremont	30.9	+/-2.0
Bronx	Castle Hill, Clason Point, and Parkchester	29.0	+/-1.8
Bronx	Bedford Park, Fordham North, and Norwood	28.7	+/-2.0
Brooklyn	Brownsville and Ocean Hill	28.5	+/-2.1
Brooklyn	East New York and Starrett City	27.7	+/-2.0

Highlighted from the NYC Poverty Index (2015-2019) we can see that precincts 40/44 have ~1/3 of the population living in poverty while precincts 73/75 have a little over 1/4 of the population living in poverty. For reference, in 2019 10.5% of the American population was living in poverty (U.S Bureau of Labor Statistics), meaning that the four precincts are on average 2.919x above the national poverty rate.

10/10 precincts with the most juvenile crime complaints are located in the top 8 poorest neighborhoods, as shown from the data provided by the NYC Poverty index. This indicates that a reason for the spike in juvenile crime in these areas is due to the conditions that the youth are living in, being that of poverty and limited opportunity.

Predicting crime paterns of youth in NYC

library(dplyr)
library(ggplot2)

# Ensure precinct column is numeric
CRIMERECENT <- CRIMERECENT %>%
  mutate(ADDR_PCT_CD = as.numeric(ADDR_PCT_CD))

# Define target precincts
target_precincts <- c(40, 44, 73, 75)

# Process youth crimes and extract hour
youth_crimes_hourly <- CRIMERECENT %>%
  filter(
    VIC_AGE_GROUP == "<18",
    ADDR_PCT_CD %in% target_precincts,
    !is.na(CMPLNT_FR_TM)
  ) %>%
  mutate(
    hour = suppressWarnings(as.numeric(substr(CMPLNT_FR_TM, 1, 2)))
  ) %>%
  filter(!is.na(hour), hour >= 0, hour <= 23) %>%
  group_by(hour) %>%
  summarise(youth_crime_count = n()) %>%
  ungroup()

# Plot
ggplot(youth_crimes_hourly, aes(x = hour, y = youth_crime_count)) +
  annotate("rect", xmin = 15, xmax = 18, ymin = 0, ymax = Inf, alpha = 0.2, fill = "red") +
  geom_line(color = "blue", linewidth = 1) +
  geom_point(color = "black", size = 2) +
  scale_x_continuous(breaks = 0:23) +
  labs(
    title = "Youth Crimes by Time of Day (with After-School Highlight)",
    x = "Hour of Day",
    y = "Youth Crime Count"
  ) +
  theme_minimal()

This graph shows the trend between juvenile crime and time of the day. As shown on the graph, crime rates peak at 15:00 and remain elevated until after 18:00. The NYC Parks Department (2019) lists after-school hours from 15:45-18:00, meaning that this period of an elevated crime rate coincides with the dismissal of children from schooling institutions. In order to curb youth crime rates, an emphasis should be made to addressing crime during this period of the day.

Afterschool location of crime in NYC

library(dplyr)
library(ggplot2)

# Step 1: Clean CRIMERECENT once with hour column
CRIMERECENT <- CRIMERECENT %>%
  mutate(
    hour = suppressWarnings(as.numeric(substr(CMPLNT_FR_TM, 1, 2)))
  )

# Step 2: Define precincts and after-school hour range
target_precincts <- c(40, 44, 73, 75)
after_school_hours <- 15:18  

# Step 3: Filter and count crimes
after_school_crimes <- CRIMERECENT %>%
  filter(
    VIC_AGE_GROUP == "<18",
    ADDR_PCT_CD %in% target_precincts,
    !is.na(CMPLNT_FR_TM),
    !is.na(PREM_TYP_DESC),
    hour %in% after_school_hours
  )

premise_counts <- after_school_crimes %>%
  group_by(PREM_TYP_DESC) %>%
  summarise(count = n()) %>%
  arrange(desc(count)) %>%
  slice_head(n = 15)

# Step 4: Plot
ggplot(premise_counts, aes(x = reorder(PREM_TYP_DESC, count), y = count)) +
  geom_bar(stat = "identity", fill = "blue") +
  coord_flip() +
  labs(
    title = "Top Premises for Youth Crime (15:45–18:00)",
    x = "Premise Type",
    y = "Crime Count"
  ) +
  theme_minimal()

As seen by the bar graph, the majority of youth crime during ‘after school hours’ occurs on the street. This is due to a variety of factors that include but are not limited to; lack of structured after-school programs, increased peer interaction without oversight, and the accessibility of streets as common gathering points for adolescents.

Limitations

• Data from this report in limited to 2019 and is five years off modern crime statistics
• Certain precincts may be represented by disproportionate amounts of youth crime due to having larger populations, this could not be accounted for as there is no reliable data on precinct population count
• Certain data was deliberately omitted due to clerical errors made while logging it into the crime database (i.e. precincts with negative numbers, crimes logged without description, etc)

Ethics Statement

Professionalism All data collected for this report was only from official government sources.

Exposing and Reviewing Methods and Findings Data was only used if it could be accessible to anyone reading the report. Everything in this report can be openly accessed by the public.

AI usage statement

ChatGPT - 4o 15/6/25 https://chatgpt.com/share/6825d0a1-72f0-8010-8079-98dd201635dc Helped me make my first graph interactive in the (Evidence) section

ChatGPT - 4o 15/6/25 https://chatgpt.com/share/6825d15f-f4c8-8010-8f44-3582770e246f Suggested how to highlight rows of my table for the NYC neighbourhood poverty index

ChatGPT - 4o 15/6/25 https://chatgpt.com/share/6825d241-e4cc-8010-a9d8-1d21e17f11f7 Suggessted how to remove error packages in my render

Acknowledgements

• R-Studio Interactive Graph Gallery: https://r-graph-gallery.com/
• Tables?: https://edstem.org/au/courses/19992/discussion/2671351
• R Coder Plots: https://r-coder.com/plot-r/

External Evidence:

• (2015-2019) NYC - Neighbourhood Poverty Index: https://www.nyc.gov/assets/opportunity/pdf/21_poverty_measure_report.pdf
• (2024.) NYC Parks Department - After School Programs: https://www.nycgovparks.org/programs/recreation/afterschool
• (2021.) Sylvana DelPrete - Poverty Correlations NYC: https://storymaps.arcgis.com/stories/0f903cebe76b461fb654271067c5fe53
• (2019.) U.S Bureau of labor statistics - Poverty Index https://www.bls.gov/opub/reports/working-poor/2019