Student details

Name: Rohit Devaraj
Student ID: s4216604

About this story

Recorded offences are not distributed evenly across Victoria. A statewide total can conceal different local trajectories and offence mixes. This visual story compares time, place, offence division, and relative change to show why one statewide figure cannot describe every community.

How to explore: Hover over marks for precise values; click legend entries to isolate series; use the Plotly toolbar to zoom or reset. All annual periods end in June. The visualisations report police-recorded offences, not an individual’s probability of experiencing crime.

1. The statewide picture is made of different offence types

From the year ending June 2017 to June 2026, recorded offences across the supplied LGA records changed by +17.1%, from 529,764 to 620,155. The largest offence division in the latest year was B Property and deception offences. The lines below show whether divisions followed the same trajectory.

2. Large offence totals are concentrated in particular LGAs

The ranking below compares the 15 LGAs with the highest official recorded-offence rates per 100,000 residents in June 2026. The tooltip also provides the recorded offence count from Table 02 where available. Population-adjusted rates are more suitable than raw counts for comparisons across LGAs, but still do not represent an individual’s personal risk.

Rates are based on resident population, not daytime visitors; city-centre and tourist areas may therefore be affected by population-denominator limitations.

3. Similar totals can conceal different offence mixes

The heatmap compares each offence division’s share of its LGA’s recorded offences for the same 15 LGAs ranked by population-adjusted rate in Chart 2. Each row sums to 100%, allowing a comparison of composition rather than absolute size. Hover over a cell to see its exact share.

4. A large current total does not necessarily mean fast growth

Each point represents an LGA. Its horizontal position shows its latest recorded-offence count, while its vertical position shows the percentage change since 2017. Points above zero increased; points below zero decreased. A percentage change in counts is not a change in population-adjusted crime risk.

5. The fastest proportional changes are not always the largest changes in volume

The final comparison focuses on percentage change between 2017 and 2026, showing the six largest increases and six largest decreases among LGAs with at least 500 recorded offences in the starting year. This threshold reduces the chance that very small starting totals dominate the ranking. The tooltip also shows the change in actual offence counts.

Blue bars indicate increases and orange bars indicate decreases. Hover to compare the percentage change with the actual change in recorded offences.

Limitations and interpretation

  • These are offences recorded by police, not all offences committed and not unique criminal incidents or victims.
  • LGA counts are not directly comparable as risk measures without appropriate population denominators.
  • Changing policing, reporting and classification practices may affect trends.
  • This report uses annual periods ending June, not calendar years.
  • Percentage changes can be unstable for small baseline counts; chart 5 uses an explicitly stated minimum baseline.
  • These data do not establish the causes of any increase or decrease.
  • Table 01 and Table 02 do not have identical statewide totals. They must not be combined into a single statewide series without reconciling geographic coverage.

Methods and reproducibility

The original official Excel workbook is imported from a local file using readxl; the R code reads Table 02 for detailed offence counts and Table 01 for official LGA rates. No data frames are hard-coded. Analysis uses offence counts aggregated by LGA and year or by offence division and year. Chart 3 divides each division’s count by its LGA’s total for the latest year. Charts 4 and 5 compare the first and latest reporting years, using (latest / first − 1) × 100. No causal model or population-risk estimate is claimed.

Source reconciliation: In the supplied workbook, the sum of Table 02 offence counts is 529,764 in 2017 and 620,155 in 2026. Table 01, excluding its regional subtotal rows, gives 529,748 and 620,140 respectively. The differences are -16 and -15. Chart 1 and Charts 3–5 consistently use Table 02; Chart 2 uses Table 01’s published rates. These sources should not be interpreted as interchangeable statewide totals.

Data, references and acknowledgements

Crime Statistics Agency. (2026). Data tables – recorded offences [Data set]. Victorian Government. https://discover.data.vic.gov.au/dataset/data-tables-recorded-offences

OpenAI. (2026). ChatGPT (GPT-6) [Large language model]. https://chatgpt.com/

Source workbook: Data_Tables_LGA_Recorded_Offences_Year_Ending_June_2026.xlsx, Tables 01 and 02, downloaded from the Crime Statistics Agency.

Note: This report uses only one official data workbook. The second uploaded visualisation workbook is retained for independent contextual checks, not combined into the calculations.

Generative AI acknowledgement

I used OpenAI’s ChatGPT (GPT-6, 2026) during the preparation of this assignment for technical guidance related to R Markdown, RStudio, Plotly visualisations, troubleshooting code, improving chart readability, and understanding data validation procedures.

ChatGPT was also used to provide suggestions relating to the presentation and organisation of the report. I reviewed the suggestions and worked with the R Markdown file and the Victorian Crime Statistics Agency datasets.

The use of ChatGPT is acknowledged in accordance with RMIT University’s guidelines on generative AI.