Is There A Crime Problem in the US?
The answer is subject to interpretation, based on personal experience and knowledge of US crime statistics.
I wanted to find the answers myself, so I spent the past 2 weeks analyzing crime data coming directly from the City of Los Angeles. There was no specific reason for choosing Los Angeles, other than the data was easily accessible and came with good quality, appropriate for analysis.
I analyzed approximately 550,000 serious crimes over the period 2020–2023. If you’re interested to observe the code, please visit the GitHub repo.
This topic is dense with information, so I will prepare you for what to expect. I came up with the following research questions, that I thought would capture the appropriate story to tell regarding crime in the US.
How Does National and Local Law Enforcement Look at Crime Data?
What are the Current Trends from the National Crime Index?
How Does the Crime In Los Angeles Compare to the National Trends?
Understanding Law Enforcement Response and the Crime Clearance Rate
Can We Use AI to Help Law Enforcement Predict Arrest Outcomes and Improve Crime Response?
Public perception of serious crimes has been identified as a top issue for the President and Congress to address in 2024. Approximately 58% of U.S. adults believe reducing crime should be a top priority, which is a 10% increase from 2021.
The cause of serious crimes is very complex and leaves a lot of uncertainty. However, it’s important not to jump to conclusions and make incorrect assumptions.
Americans tend to believe crime is up, even when official data shows it is down. In 23 of 27 Gallup surveys conducted since 1993, at least 60% of U.S. adults have said there is more crime nationally than there was the year before, despite the downward trend in crime rates during most of that period.
How Does National and Local Law Enforcement Look at Crime Data?
The U.S. Department of Justice administers two statistical programs to measure the magnitude, nature, and impact of crime in the nation: Listed as follows:
- Federal Bureau of Investigations (FBI) Uniform Crime Reporting Program Summary Reporting System (UCR SRS) and
- Bureau of Justice Statistics’ (BJS) National Crime Victimization Survey (NCVS).
It’s important to understand the limitations of these programs.
The Uniform Crime Reporting (UCR) Program divides offenses into two groups, Part I and Part II crimes. Each month, participating law enforcement agencies submit information on the number of Part I offenses. Contributors only provide arrest data for Part II crimes. The data does not include drug or hate crimes.
Part I offenses make up the National Crime Index because they are serious crimes, they occur with regularity in all areas of the country, and they are likely to be reported to police. These crimes are broken down into two categories, Violent Crime and Property Crime, which can be broken down further, listed as follows:
Violent Crime — Homicide, Aggravated Assault, Robbery, Rape
Property Crime — Larceny/Theft, Burglary, Motor Vehicle Theft, Arson
BJS, tracks crime by fielding a large annual survey of Americans ages 12 and older and asking them whether they were the victim of certain types of crime in the past six months. One advantage of this approach is that it captures both reported and unreported crimes. A disadvantage is that BJS is a survey that deals with crimes after the fact, so it does not track homicide.
The FBI is transitioning to a new data collection system to improve data quality, known as the National Incident-Based Reporting System (NIBRS). This system collects data on a much larger set of crimes, including information on victims, known offenders, relationships between victims and offenders, arrestees, and property involved in crimes.
In 2022 the FBI acquired crime data that provided coverage for 95% of the population, the most complete set of crime data to date. We will analyze crime before, durring and after 2022 for this analysis.
What are the Current Trends from the National Crime Index?
Reported Violent Crime Rate in the US 1990 — 2022
In 2022, there were roughly 1.23 million violent crimes reported to the FBI in the United States. This number can be compared to the total number of property crimes, roughly 6.51 million that year.
- 893,980 aggravated assaults, making this offense the most common of violent crime offenses.
- Murder and non-negligent manslaughter recorded a 2022 estimated nationwide decrease of 6.1% compared to the previous year.
- In 2022, the estimated number of offenses in the revised rape category saw an estimated 5.4% decrease.
- Aggravated assault in 2022 decreased an estimated 1.1% in 2022.
- Robbery showed an estimated increase of 1.3% nationally.
- Unrelated to this study, but should be mentioned — 19,592 people were killed by a gun in 2022, the second-highest gun homicide rate since 1995.
Reported Property Crime Rate in the US 1985–2022
- In 2022, the national property crime rate increased for the first time in over two decades, rising 6.7%
- Nearly 8% jump in larcenies/theft, a category that includes most forms of theft, including shoplifting. It is the first such increase in years. But even with these increases, the rate of larcenies remains below both 2019 and 2020 levels.
- More recent data seems to confirm the impression that larcenies appear to be falling in 2023.
How Does the Crime In Los Angeles Compare to the National Trends?
There’s an escalating issue with Property Crimes, which continues to escalate beyond 2022.
Violent crimes were escalating until 2023 and then began to decline in 2023.
Let’s take a closer look at Violent Crime.
Let’s take a closer look at the numbers.
- 2020 showed increases in all violent crimes.
- Homicides made a 14% jump from 2020–2021
- 8% jump in aggravated assault, during the same period.
- 2022 drew increases in Robbery and aggravated assault, with decreases in homicide and rape.
- Finally 2023 shows noticeable drop in most violent crimes, particularly homicide at -16%. However, there is still a big problem with Aggravated Assault, as the increase in crime is mostly left unchanged.
- In California, the 2022 property crime rate was up 5.9% from 2021,
- Up by 8.7% from 2020, when it had reached the lowest level observed since 1960.
- Property crime is now slightly above the pre-pandemic 2019 level.
- Auto theft continued its increase since 2020, up by 1.6%
- In 2023, 202,802 vehicles were stolen, at an estimated total value of $1.8 billion.
- Approximately 51.7 percent of all vehicle thefts occurred in Southern California, with 60 percent occurring in Los Angeles.
- Theft also increased in 2022, by 7.3%. For the first time since 2012, burglary increased in 2022.
Understanding Law Enforcement Response and the Crime Clearance Rate
Most violent and property crimes in the U.S. are not reported to law enforcement, and most of the crimes that are reported are not solved.
In its annual survey, BJS found that in 2022, only 41.5% of violent crimes and 31.8% of household property crimes were reported to authorities.
An agency’s clearance rate is the share of Part I violent and property crimes that are considered solved after law enforcement makes an arrest.
In 2022, police nationwide cleared 36.7% of violent crimes that were reported to them and 12.1% of the property crimes that came to their attention.
Crime Clearance Rates in Los Angeles
California statewide overall “clearance rate” — crimes resulting in an arrest being made — was 13.2 percent in 2022. The clearance rate for property crimes was even lower, at 7.2 percent, while violent crimes were cleared by law enforcement 41 percent of the time.
According to LAPD crime data, the average clearance rate is 24% for all crimes, which shows a 11% improvement compared to the state.
- Citywide Violent Crimes showed a 3.2% decrease in 2023
- Citywide Part 1 Crimes showed a 1.9% increase from 2022
- In 2023, there were 327 homicides, which was a 17% reduction in homicides as compared to 2022
- The homicide clearance rate City-wide average was at 76%
The topic of clearance rates is a bit controversial. While funding for LAPD is said to approach approximately $25 billion, crime clearance rates seem to be getting worse, declining 41% over the past 30 years. There are reports that the police force participation is much lower than in the 1990s.
However, the cause of decline in clearance rates goes way beyond the scope of this analysis.
Can We Use AI to Help Law Enforcement Predict Arrest Outcomes and Improve Crime Clearance Rates?
I believe the answer is YES! You can call me naive, I don’t care. I really believe Data Science can be a key tool in improving crime clearance rates.
The average number of crimes that come into the LAPD is about 370 per day and comes with a lot of volatility on a daily basis, which I can only imagine can be very stressful.
I can’t help but think, with all the data available, what if we could predict an arrest outcome as soon as it’s reported? How would that change resource allocation and prioritization, if we could predict outcomes with a high level of accuracy? How would you be able to ignore that data? How would we change our approach to clearing the crime? How would these tools effect public perception and collaboration with law enforcement? If more citizens were aware of these tools, would it motivate them to report more crime?
I ran a quick experiment with the LAPD crime Dataset, training a machine learning model to predict arrest outcomes for 549,861 Part 1 crimes.
- The model was trained on 384,902 Part 1 crimes
- The model was tested on 109,973 Part 1 crimes
- When testing for accuracy, the model accurately predicted arrest outcome 89% of time on over 100,000 crimes
If you look at the confusion matrix, we’ll start with the numbers where the model predicted correctly.
The number in yellow (92,710) represents true negative outcomes. This means that from the 109,973 crimes that the model saw, it predicted ‘no arrest’ with 90% accuracy. The number on the bottom right (4,671) are the true positives, the crimes that resulted in arrest, predicted with 70% accuracy.
Now to the errors.
The bottom left (10,546) represents the false positives, the top right represent the false negative (2,046). Roughly 13,000 crimes were predicted incorrectly.
What does this all mean?
In order for the model to provide value to law enforcement and it’s citizens, it needs to be more accurate. There are so many things we can do to improve the model, but it requires time and engineering resources.
With open source tools, the data pipelines and ML model was built in 2 days. With more time and more resources, I believe a more accurate model can be built ready for production.
Conclusions on National Crime Statistics and Crime Status in Los Angeles
Los Angeles is experiencing some serious crime problems, that have caused the mayor to act and present the following statistics on 2024 Q1 crimes and clearance rates.
How serious is the problem? You would have to compare Los Angeles to other Cities on a granular level, which goes beyond the scope of this analysis.
What have we learned?
National headlines don’t always translate locally. Crime is complex and there is no absolute cause of crime that is obvious that we can analyze or respond to. However, the common message that comes from law enforcement and think tanks is that population density, quality of life, mental health and economic conditions have a big influence on crime.
Giving people opportunity to receive high quality healthcare services, improve personal safety and economic opportunity, living safe and secure lives, will make for a better society.
