A Decade of Crime in London: Insights from Spatial-Temporal Data

7 min readDec 17, 2023

London, a city known for its rich history, also has a story to tell about crime over the past decade. In this article, we analyze crime data from 2010 to 2023 to understand how it has changed across different boroughs. The aim of this article is to use data analysis and visualisations to uncover the trends and shifts in London’s crime scene.

Data

This analysis is based on data from the ‘Recorded Crime Summary’ dataset, available on the London Datastore, which encompasses detailed records of crime from April 2010 to September 2023. It comprehensively covers ten major categories of crime, each further divided into various minor subcategories, providing a granular view of London’s crime landscape over the years. The data was from the Metropolitan Police and thus does not contain data on the City of London, which has its own police force.

In this article, we look at the distribution of crime in London, crime evolution from 2012 to 2023 and spatial-temporal trends.

Looking at Major Crime counts

This section aims to answer the following questions.

What are the major crime groups?
And how are they distributed?

The dataset is grouped by Major Crime categories. Each Major Crime category has subcategories called Minor Crime categories. For example, a major category, Drug Offences, is broken down into two Minor Categories: Drug Trafficking and Possession of Drugs.

This section examines the number of incidents for each Major Crime category, detailing the counts within their respective Minor Categories.

Fig 1 — Total Incidents of Major Crime Types: 2010–2023

Fig 1 indicates that Theft and Violence Against the Person are the highest reported crimes in London from 2010 to 2023. Historical Fraud and Forgery and Possession of Weapons are among the lowest during this period.

Fig 2 — Crime Statistics Overview: 2010–2023

Fig 2 breaks down each Major Category into its Minor Categories. We see for Drug Offences, Possession of Drugs is significantly higher than Drug Trafficking, which could imply a larger focus or success in apprehending individuals for possession compared to trafficking, or it may reflect a wider and more border issue of drug use within London.

In the category of Violence Against the Person, incidents classified as Violence without injury significantly outnumber those resulting in injury, suggesting that a substantial portion of these crimes involve intimidation, threats, or minor altercations that do not lead to physical harm. This could reflect a prevalence of less severe but more frequent forms of violent behaviour, which, while not leading to physical injury, still contribute to a sense of insecurity and can have a lasting impact on the community.

Evolution of Crimes

This section aims to answer the following questions.

How has crime in London evolved over the past decade?
Which crimes show an increase and which show a decrease? Do all crimes follow a similar trend?

Fig 3 — Monthly Crime Distribution (Left) and Time Series of Minor Categories (Right) in London 2010–2023

Two types of time series plots are displayed in Fig 3. To the left, we have a monthly distribution of Major Crime categories. The darker the colour, the higher the count. To the right, we have the evolution of each Minor Crime category.

In the figure to the left, the x-axis represents the year, and the y-axis represents the month. The colour represents the crime count. There's quite a lot of variation happening between Major Crime categories. Some are increasing while others are decreasing.

For instance, on the bottom left for Sexual Crimes, we see a lighter shade around 2010–2012, which gets darker as the years progress, suggesting an increase in Sexual Crimes. These changes could be attributed to a genuine rise in incidents or could be from improved victim support leading to increased reporting. We see from the right figure that Sexual Crimes are broken down into two Minor Categories. A more general category called Other Sexual Offences and Rape. Both seem to follow the same trend, with their highest peaks in 2022.

Looking at Violence Against the Person (bottom right), there is a clear increase over the years, with the rise starting in 2012 and continuing until 2023. The highest point is in 2022. Violence without injury, which we saw, made up the most of this category, increased the most. In comparison, violence with injury has shown a steady trend over the same time period.

Most Dangerous Boroughs

In this section, we aim to answer two questions.

Which Borough takes the crown for being the most dangerous?
What are the most common crimes in each area of London?

Slightly different from the above sections, where we looked at the crime count in London. This section looks at the crime rate per 1000 people in each Borough. This standardizes the measurement of crime across areas with differing populations, allowing a fairer comparison between Boroughs of different sizes. This metric is calculated as follows:

1000 * (total_crimes / population) # Crimes per 1000 people

First, which Borough takes the crown for being the most dangerous?

Fig 4 — Average Yearly Crime Rate between 2010 and 2023

The bar chart in fig-4 reflects the average yearly crime rate per 1000 people from 2010 to 2023.

Westminster 👑 , which tops the chart, is a prime example where its status as a central borough with key attractions, government buildings, and a high footfall of tourists and locals alike, correlates with higher instances of crime, particularly theft and public order offences. Camden, similarly, with its vibrant markets and nightlife, sees elevated crime rates, possibly due to the large crowds it draws, creating opportunities for petty crimes.

Fig 5 — Average Annual Crime Rates per 100 Residents by Borough and Crime Type (2010–2023)

Looking at the figures in Figure 5, we can see the most common crimes in each borough in London. Westminster, Camden and Islington top almost every chart.

Crime Heat Maps

The choropleth maps in Figure 6 depict spatial temporal crime trends in London. The cutpoints used are the [0.25, 0.5, 0.75, 0.9, 1] quantiles. Each map represents the spatial distribution of crime for that year, and each map uses the same quantiles so we can easily compare across different years. Darker regions are locations with higher crime rates per 1000 people.

Central London boroughs like Westminster, Camden and Islington see elevated crime rates across all years. Westminster sees around 404 crimes per 1000 people, while the other central boroughs have more than 130 crimes per 1000 people.
Decrease in crime rate around West London. Evident through the change in colour. Colours lightened from 2013 to 2016 in boroughs like Hounslow, Ealing, and Brent. Figures went from 103–128 crimes per 1000 people to between 80–103 crimes per 1000 people in this period. During the same period, crime rates in East London, in boroughs like Newham, Tower Hamlets and Hackney, were 103–128 crimes per 1000 people.
Seems as though since 2017 crime has spread outwards from inner London in all directions. We see regions that were previously white in colour, indicating a low crime rate of around 49–66 crimes per 1000 people shift to 66–80 crimes per 1000 people. Noticeable in boroughs like Kingston upon Thames, Merton, Bromley and Barnet.

Fig 6 — Spatial-Temporal Distribution of All Crimes Per 1000 People

Please see the appendix section for spatial-temporal plots of different types of crimes.

Next Steps

The next steps for this project would include modelling and trying to predict the risk of crime using machine learning. This was actually part of my dissertation paper, where Bayesian Spatial-Temporal Models were used to model the risk of violent crime in London.

It would also be interesting to find relationships between different crime types. For example how a drug abbuser might resort to theft to finance their drug habits.