MA304 Assignment – Crime & Weather in Colchester (2025)

1 Introduction

Colchester, one of the oldest recorded towns in Britain, combines a historic core with a growing university population and suburban residential areas. This report analyses street-level crime incidents recorded by Essex Police across Colchester during 2025, alongside daily weather observations from a nearby meteorological station. As a Level 7 submission, the report includes interactive elements built with Plotly and Leaflet; these are embedded directly in the HTML output and can be explored by hovering, zooming, and clicking within the relevant figures (Figures 9, 13, and 16). The objective is threefold: first, to characterise the volume, type, and spatial distribution of crime in Colchester; second, to explore temporal patterns and their alignment with seasonal weather variation; and third, to consider the social and ethical dimensions of using such data for policy insight.

The crime data are drawn from the UK Police open-data portal (data.police.uk), which publishes monthly street-level records with obfuscated coordinates snapped to the nearest street or landmark. Weather data come from OGIMET station 3590 (near Colchester), providing daily summaries of temperature, precipitation, humidity, wind, and sunshine hours. By merging these two sources at the monthly level, we can examine whether environmental conditions coincide with shifts in criminal activity — whilst remaining mindful that correlation does not imply causation, and that a single year of data limits the strength of any inference.

The analysis is structured as follows: Section 2 details the data preparation and recoding steps; Section 3 presents categorical summaries through cross-tabulations; Section 4 provides a range of static and interactive visualisations; Section 5 examines temporal and spatial patterns; Section 6 investigates multivariate relationships between weather and crime; Section 7 discusses vulnerability and equity; Section 8 addresses ethical considerations; and Section 9 summarises the main findings.

2 Data Preparation

2.1 Crime Data

The crime dataset is read with manually specified column names, since the raw file includes a header row that we replace for consistency. Coordinates are coerced to numeric and rows with missing or out-of-bounds values are removed. The date field (formatted as YYYY-MM) is converted to a proper Date object by appending -01, and derived temporal variables (month name, month number, season) are created. Three new grouping factors are then constructed: crime_group collapses the detailed crime categories into five broader types; loc_grp classifies locations by their street-name keywords; and outcome_grp simplifies the policing-outcome text into four summary categories. Finally, the data are deduplicated on the incident identifier and converted to a spatial sf object for mapping.

After cleaning, the crime dataset comprises 5956 incidents spanning 12 months, across 14 original categories collapsed into 5 groups.

2.2 Weather Data

The weather file is read with its own header row. Column names are cleaned to lower snake-case and key variables are renamed to shorter, descriptive labels. Numeric coercion, filtering to the year 2025, and derivation of month and season fields follow the same logic as the crime data. A monthly weather summary is then produced, and the two datasets are merged on month number.

The weather dataset covers 365 daily records for 2025. The merged monthly table contains 12 rows — one per calendar month — linking aggregate crime counts with mean temperature, total precipitation, mean humidity, mean wind speed, and total sun hours.

3 Categorical Analysis

This section examines the cross-tabulated relationships between crime types, locations, temporal periods, and policing outcomes. Three two-way tables are presented, each providing a different lens on how crime is distributed in Colchester.

3.1 Crime Group by Month

Table 1: Crime incidents by group and month (2025)

crime_group	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sept	Oct	Nov	Dec	Total
Violent	166	188	179	191	244	232	224	244	184	252	214	201	2,519
Property	143	171	158	186	185	188	183	192	165	193	148	144	2,056
Public Order / ASB	65	85	86	106	131	85	84	95	101	64	88	52	1,042
Drugs & Weapons	24	16	20	22	28	22	10	23	18	15	23	29	250
Other	10	5	4	13	10	4	9	8	4	8	4	10	89

Table 1 reveals that Violent crime consistently accounts for the largest share of incidents across every month, with notable peaks during the warmer months. Property crime forms the second-largest category, while Public Order / ASB incidents show a broadly stable pattern with slight summer increases. Drugs & Weapons and Other categories remain relatively low throughout the year.

3.2 Crime Group by Location

Table 2: Crime incidents by group and location type

crime_group	Retail	Nightlife	Open Space	Residential Street	Other	Total
Violent	178	117	222	1,554	448	2,519
Property	608	60	125	1,018	245	2,056
Public Order / ASB	128	43	76	621	174	1,042
Drugs & Weapons	19	6	27	132	66	250
Other	7	4	9	52	17	89

Table 2 highlights that Residential Streets host the highest volume of crime overall, reflecting the spatial distribution of population and housing. Retail areas are a notable secondary hotspot — particularly for Property crime (shoplifting, theft) — while Nightlife locations are disproportionately associated with Violent and Public Order / ASB incidents relative to their smaller geographic footprint.

3.3 Outcome by Crime Group

Table 3: Policing outcomes by crime group

outcome_grp	Public Order / ASB	Violent	Property	Drugs & Weapons	Other	Total
No Outcome (ASB)	590	0	0	0	0	590
No Further Action	328	1,924	1,667	45	44	4,008
Formal Action	70	270	191	148	23	702
Ongoing / Other	54	325	198	57	22	656

Table 3 shows that the majority of incidents result in either “No Further Action” (investigation complete but no suspect identified, or unable to prosecute) or “No Outcome (ASB)” which applies to anti-social behaviour records that carry no formal policing outcome by design. “Formal Action” — including cautions, court outcomes, and local resolutions — makes up a comparatively small share, underscoring the well-documented challenge of achieving successful prosecutions in high-volume, low-severity crime categories. “Ongoing / Other” includes cases still under investigation at the time of data extraction.

4 Visualisations

This section presents a series of static plots that explore crime volume, distribution across categories and locations, and the shape of weather variables across seasons. Each figure is designed to highlight a specific pattern or contrast.

4.1 Bar Chart: Total Crimes by Group

Figure 1: Total crime counts by group (2025)

Figure 1 confirms that Violent crime dominates the dataset, followed by Property crime. Public Order / ASB ranks third, while Drugs & Weapons and Other crime types are relatively infrequent. This hierarchy is consistent with national patterns for a mid-size English town, where violence against the person has been the single largest recorded category in recent years.

4.2 Dot Plot: Monthly Crime Counts

Figure 2: Total incidents per month (2025)

Figure 2 uses a Cleveland dot plot to display total crime counts by month. The visualisation makes it straightforward to compare months and identify any seasonal ramp-up. If a summer peak is visible, it would align with the well-established finding that warmer weather brings more outdoor activity and, consequently, more opportunity for certain crime types.

4.3 Pie Chart: Outcome Proportions

Figure 3: Policing outcome distribution

Figure 3 illustrates the proportion of each outcome group. The dominance of “No Further Action” highlights the systemic difficulty of resolving street-level incidents, particularly property and public-order offences where evidence is often limited. The “No Outcome (ASB)” slice reflects anti-social behaviour records, which by design do not receive a formal policing outcome.

4.4 Histogram: Average Daily Crime per Month

Figure 4: Distribution of average daily crime rate across months

Figure 4 provides a histogram of the average daily crime rate computed for each month (monthly total divided by the number of days in that month). This normalisation removes the bias introduced by differing month lengths. The distribution gives a sense of how tightly clustered monthly crime rates are, or whether particular months stand out as unusually high or low.

4.5 Density Plots: Temperature by Season

Figure 5: Distribution of average daily temperature by season

Figure 5 uses overlaid density curves to show how daily average temperatures are distributed within each season. The clear separation between Winter and Summer confirms the expected seasonal gradient in Colchester, while Spring and Autumn display broader, overlapping distributions, reflecting their transitional character. These temperature distributions will be important when interpreting seasonal crime patterns later in the report.

4.6 Box and Violin Plots: Monthly Crime by Group

Figure 6: Monthly crime distribution by group (box + violin)

Figure 6 combines violin and box plots to reveal both the shape of the distribution and its summary statistics. Violent crime shows both the highest median and the widest spread, meaning it fluctuates more month to month than other groups. Property crime is the next most variable. The narrower violins for Drugs & Weapons and Other indicate relative stability, though their lower medians mean individual monthly spikes could still be proportionally significant.

4.7 Sina Plot: Location Groups Across Months

Figure 7: Incident distribution by location group across months

Figure 7 uses a sina plot — a jittered strip chart that respects the local density of points — to show how incidents within each location group are distributed across the months. Retail and Residential Street locations appear populated throughout the year, while Nightlife incidents may cluster in particular months. Open Space incidents could be expected to increase during warmer months when parks and public areas receive more foot traffic.

5 Temporal and Spatial Analysis

This section explores how crime evolves over time and where it concentrates geographically. Static and interactive time-series plots reveal trends and seasonal shifts, while maps display the spatial distribution of incidents across Colchester.

5.1 Static Time Series with LOESS Smoother

Figure 8: Monthly crime trend with LOESS smoother

Figure 8 plots total monthly crime counts with a LOESS (locally estimated scatterplot smoothing) curve superimposed. The smoother helps to identify the underlying trend without being misled by month-to-month noise. Any upward trajectory during the spring-to-summer transition — and subsequent decline in autumn — would corroborate the seasonal hypothesis explored throughout this report.

5.2 Interactive Time Series: Crime and Temperature

The interactive plot below uses Plotly to overlay monthly crime counts and average monthly temperature on a dual-axis chart. Hovering over data points reveals the exact values. This is one of the Level 7 interactive elements included in the report.

Figure 9: Interactive dual-axis — crime counts and temperature

The interactive dual-axis chart reveals the degree to which monthly crime counts and average temperature track one another. Months where the two lines rise and fall in tandem suggest a shared seasonal driver — likely increased outdoor activity and social mixing in warmer weather — although this visual overlap does not constitute causal evidence.

5.3 Seasonal Heatmap

Figure 10: Seasonal heatmap of crime group by month

Figure 10 presents a tile heatmap in which colour intensity encodes the number of incidents. The heatmap makes it easy to scan both across months (for a given crime type) and across crime types (within a given month). Violent crime stands out as the consistently darkest row, while the seasonal gradient is visible horizontally for most groups.

5.4 Ridgeline Plot: Crime Types Across Months

Figure 11: Ridgeline plot of crime types across months

Figure 11 uses ridgeline (joy) plots to show how each crime group is distributed across the months. Unlike the heatmap, this format emphasises the shape of the distribution: whether incidents are concentrated in certain months or spread evenly. A flatter ridge indicates a more uniform distribution; a pronounced peak suggests seasonality.

5.5 Static Spatial Map

Figure 12: Spatial scatter of crime incidents in Colchester

Figure 12 plots each aggregated location as a point on a Cartesian scatter map. Larger points indicate higher local incident counts. The clustering of large, multi-coloured points near the town centre reflects the concentration of retail, nightlife, and transit hubs. Peripheral areas tend to have smaller, sparser points, consistent with lower population density and fewer commercial premises.

5.6 Interactive Leaflet Map

The interactive Leaflet map below provides a heatmap layer showing crime density, with individual markers that display popups containing the crime category, street name, date, and policing outcome. Users can zoom, pan, and toggle layers.

Figure 13: Interactive Leaflet heatmap of crime density

The Leaflet map enables granular exploration of spatial patterns. The heatmap layer immediately highlights the town-centre corridor as the primary crime hotspot, with secondary clusters near the railway station and major retail areas. Clicking individual markers reveals the specific incident details, allowing users to drill into the data at street level.

6 Multivariate Analysis

Having examined crime and weather separately, this section investigates their joint variation. Scatter plots, an interactive bubble chart, pair plots, and a correlation matrix heatmap are used to assess whether monthly weather conditions are statistically associated with crime volume.

6.1 Scatter: Temperature vs Crime

Figure 14: Monthly average temperature vs crime count

Figure 14 plots monthly crime counts against average temperature. The linear fit (red) provides a summary of the overall direction of association. If the relationship is positive — more crime in warmer months — it is consistent with routine activity theory, which posits that increased ambient population and reduced guardianship (people away from home) raise crime opportunity.

6.2 Scatter: Precipitation vs Crime

Figure 15: Monthly total precipitation vs crime count

Figure 15 examines whether rainfall deters crime. Theory suggests that heavy precipitation reduces outdoor activity and thus crime opportunity. The direction and strength of the linear fit will indicate whether Colchester’s 2025 data support this hypothesis. Any inverse relationship should be interpreted cautiously, given the confounding role of seasonality (summer months tend to be both warmer and drier).

6.3 Interactive Bubble Chart

Figure 16: Interactive bubble — temperature, crime, precipitation, humidity

The interactive bubble chart encodes four variables simultaneously: temperature on the x-axis, crime count on the y-axis, precipitation as bubble size, and humidity as colour. Hovering over a bubble reveals the month label and exact values. This multivariate view enables the reader to assess, at a glance, whether months with high temperature and low precipitation tend to coincide with higher crime, or whether the relationship is more nuanced.

6.4 Pair Plot (GGally)

Figure 17: Pair plot of crime and weather variables

Figure 17 provides a comprehensive overview of all pairwise relationships between monthly crime counts and weather variables. The upper triangle displays Pearson correlation coefficients; the diagonal shows the univariate density of each variable; and the lower triangle provides scatter plots with LOESS smoothers. This format is particularly useful for identifying unexpected correlations (e.g., between humidity and crime) or confirming expected ones (temperature and sunshine).

6.5 Correlation Matrix Heatmap

Figure 18: Correlation matrix heatmap with hierarchical ordering

Figure 18 presents the same correlation structure as a colour-coded heatmap with hierarchical clustering. Variables that are most similar in their correlation pattern are grouped together. The rectangular outlines highlight natural clusters. This visualisation provides a concise summary of which weather variables are most strongly — and most weakly — associated with monthly crime volume.

7 Vulnerability and Equity

The spatial and categorical patterns revealed above have implications that extend beyond academic description. Understanding who is most affected by crime, and where, is essential for equitable policy-making.

The concentration of incidents in the town centre — particularly around retail areas, the nightclub district, and transport hubs — means that individuals who live, work, or commute through these zones bear a disproportionate burden. Retail and service-sector workers, who are often on lower incomes and less able to choose alternative routes or hours, are especially exposed to shoplifting, public order offences, and violent incidents.

Anti-social behaviour, while sometimes considered a lower-severity category, can have a significant cumulative impact on wellbeing and quality of life. The data show that ASB is spread across residential and open-space locations, suggesting that everyday environments — parks, streets, housing estates — can be sites of sustained disorder. For vulnerable populations, including the elderly, people with disabilities, and families with young children, persistent ASB can erode the sense of safety that underpins community cohesion.

The outcome data further underscore equity concerns. The large proportion of “No Further Action” outcomes suggests that many victims, particularly those of property crime and lower-level violence, do not see their cases resolved. This can reinforce feelings of powerlessness and distrust in institutions, particularly in communities that are already marginalised.

Colchester’s growing university population adds another dimension. Students — many of whom are new to the area, living in unfamiliar settings, and socialising in the nightclub corridor — may be disproportionately affected by violent and public-order incidents but less likely to report them or engage with the criminal justice process.

Addressing these disparities would require not only more granular data (e.g., linking crime records with the Index of Multiple Deprivation) but also qualitative engagement with affected communities to understand their lived experience of safety and insecurity.

8 Ethical Considerations

8.1 Privacy and Coordinate Obfuscation

The UK Police open-data portal deliberately snaps crime locations to the nearest street or landmark rather than publishing exact addresses. This is an essential privacy safeguard: even approximate coordinates could stigmatise specific households or buildings if published at higher resolution. However, this obfuscation introduces spatial uncertainty — incidents attributed to a shopping area may have occurred within a radius of several hundred metres. Analysts should avoid over-interpreting localised clusters, since they may partly reflect the snapping algorithm rather than genuine hotspots.

8.2 Under-Reporting and Policing Intensity

Recorded crime statistics are a product of both criminal activity and the apparatus that records it. Certain crime types — domestic violence, sexual offences, hate crime — are systematically under-reported, meaning that the observed distribution in the data does not fully capture the true distribution of harm. Conversely, areas with higher police visibility (such as the town centre) may have inflated recorded rates simply because more officers are present to detect and record incidents. The high density near Colchester’s police station likely reflects this dynamic as much as it reflects genuine crime concentration.

8.3 Limitations of Single-Station Weather Data

Using a single weather station to represent conditions across the entire Colchester area is a simplification. Microclimatic variation — differences in wind exposure, urban heat island effects, or localised rainfall — can be meaningful at the street level but is invisible in station-level averages. Moreover, daily averages smooth over intra-day variation; a warm afternoon may be relevant to crime opportunity, but a cold morning could suppress it, and both contribute to the same daily mean.

8.4 Single-Year Temporal Limitation

With only one year of data (2025), it is impossible to distinguish genuine seasonal effects from year-specific anomalies. An unusually warm winter, a major public event, or a change in policing strategy could all produce patterns that would not replicate in other years. Multi-year data would be needed to estimate robust seasonal effects and to control for long-term trends.

9 Conclusion

This analysis of street-level crime in Colchester during 2025 reveals several key patterns. Violent crime is the most prevalent category, followed by property offences and public-order incidents. Spatially, crime is concentrated in the town centre — particularly around retail areas, the nightlife corridor, and the railway station — with lower levels in suburban residential zones. Temporally, the data suggest a seasonal pattern consistent with routine activity theory: crime volume tends to be higher in warmer months when outdoor activity increases.

The multivariate analysis indicates a positive association between temperature and monthly crime counts, with some evidence that precipitation may act as a deterrent. However, these relationships are based on only 12 monthly observations and should be treated as exploratory rather than confirmatory. The correlation and pair-plot analyses highlight which weather variables track crime most closely and which do not.

From an ethical standpoint, the analysis underscores the importance of treating crime statistics as a partial and potentially biased representation of reality. Coordinate obfuscation, under-reporting, policing intensity, and single-source weather data all limit the conclusions that can be drawn. Future extensions of this work might incorporate multi-year data to strengthen seasonal estimates, integrate deprivation indices (IMD) to assess whether crime disproportionately affects disadvantaged neighbourhoods, or explore university term-time effects on the nightlife corridor.

10 References

UK Police Open Data Portal. Available at: https://data.police.uk/.

OGIMET Synoptic Data. Available at: https://www.ogimet.com/.

Bczernecki, B. (2023). R climate package (OGIMET interface). Available at: https://bczernecki.github.io/climate/.

Tierney, N. (2020). UK Police API (R package). Available at: https://ukpolice.njtierney.com/.

Wickham, H. et al. (2019). Welcome to the Tidyverse. Journal of Open Source Software, 4(43), 1686.

Cohen, L. E. and Felson, M. (1979). Social Change and Crime Rate Trends: A Routine Activity Approach. American Sociological Review, 44(4), 588–608.

Field, S. (1992). The Effect of Temperature on Crime. The British Journal of Criminology, 32(3), 340–351.

Cohn, E. G. and Rotton, J. (2000). Weather, Seasonal Trends, and Property Crimes in Minneapolis, 1987–1988. Journal of Environmental Psychology, 20(3), 257–272.