UK Census project

This project presents a comprehensive analysis of a fictional UK town using synthetically generated census data. Modelled after the 1881 UK census, the dataset was created using Python's Faker library and consists of 8,485 individual records. The goal is to emulate how census data can be used to guide public policy, investment, and social planning decisions through data cleaning, statistical analysis, and visual storytelling.

Data Preparation & Cleaning

Data cleaning addressed missing values, format inconsistencies, and incorrect inputs. Key steps included:

• Dropping irrelevant index columns.
• Converting age data from text/float to integer.
• Inferring missing relationships and occupations using household context.
• Normalising gender, marital status, religion, and occupation fields.
• Categorising occupations into employment types for easier analysis.
• Removing or adjusting implausible entries (e.g. underage heads of household).

Demographics

The town is dominated by a middle-aged population (20-60 years), with females slightly outnumbering males. This indicates a potential ageing population crisis in the future, particularly in health and social care.

Employment Trends

Employment levels are strong, with most residents categorised as employed or students. A derived employment category field helped segment the population into employed, unemployed, students, children, and retirees. Most economically active individuals are aged 20-60.

Religion

Roughly 33% of residents have no religion. Christianity remains the most declared religion (~23%). Other affiliations, such as Islam, Sikhism, or Jediism, were minimal and normalised in the data.

Marital Status by Age

Marriage, divorce, and widowhood span across all age groups, with a larger number of widowed individuals in older age bands. There are more divorced females than males, likely indicating gendered social trends.

Commuting and Migration

Around 43% of residents are estimated to commute out of the town regularly—mostly university students and working professionals. Estimated immigration and emigration rates (46.69 and 21.9 per 1,000 respectively) suggest a modest population growth of 19.2 per 1,000 households.

Housing Occupancy

Analysis of occupancy across 2,819 households showed several homes under-utilised, indicating low demand for new housing development. Instead, improving use of existing housing stock is recommended.

Strategic Recommendations

• Create more local employment opportunities (e.g. manufacturing).
• Improve transport infrastructure (consider a train station).
• Expand or prepare healthcare services for ageing residents.
• Avoid building new houses; repurpose underutilised ones.
• Build one additional Christian place of worship to reflect religious spread.
• Support community education and training for the unemployed.

Conclusion

This project demonstrates how synthetic data can simulate real-world census analysis. With proper data cleaning and visualisation, valuable insights were derived to guide urban planning, social infrastructure, and investment decisions.