What forecasting is not

Predictive policing

Time: hours/days into future

Geography: streets or grid cells

Horizon scanning

Time: (many) years into future

Geography: national/international

Crime forecasting

Time: weeks/months into future

Geography: national/force/district

Forecasting uses data on the frequency of events …

… to estimate how many events will occur in the future

Forecasting breaks down time series into components

Time series = trend +

Time series = trend + season +

Time series = trend + season + other

Many repeating and one-off events can influence time series

Public holidays

Sporting events, festivals, etc.

Emergencies

Forecasting relies on structural stability

But …

… new technologies change crime patterns

… policing itself changes

… unexpected events happen

Forecasts should be expressed as:

If the factors underlying crime remain broadly the same, we expect …

How forecasting could potentially be useful to policing

Scenario 1

A police commander preparing a budget request wants to know how many crimes are likely to occur in a city each month for the next three years

Scenario 2

The officer in charge of shift assignments wants to know how many crimes are likely to occur each day for the next three months.

Scenario 3

A senior detective wants to know how many serious assaults are likely to occur in a district each month for the next year.

Crime analysts

  • must master lots of skills
  • have limited time
  • often have limited software access

⇒ need forecasting to be semi-automatic

The current study

compared 10 forecasting methods

using crime data from 12 US cities

with 36 tests of each scenario in each city

⇒ 1,296 tests of the performance of each forecasting method

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

combined (‘ensemble’) model = mean of ETS, linear model, seasonal naïve and STL forecasts

Conclusion

  • Reasonably accurate forecasting is possible
  • A combination of forecasts from multiple methods is typically most accurate
  • Some specific methods (e.g. FASSTER) are wildly often inaccurate

Remember

Forecasts are only useful as long as there is structural stability

these slides

these slides:
lesscrime.info/talk/esc-2025/

article pre-print:
doi.org/p4vn

questions:
matthew.ashby@ucl.ac.uk

free browser-based tool for forecasting

coming when I have time to finish coding it