Skip to contents

Crime data for Canadian provinces and territories, showing mischief offenses. This dataset is adapted from the original Bayesian Surprise paper's Canada example and is useful for exploring base-rate effects: Ontario and Quebec dominate raw counts because of population, while model-based surprise scores ask which provinces are unusual relative to the chosen model space.

Usage

canada_mischief

Format

A data frame with 13 rows and 6 variables:

name

Province or territory name

population

Population count

mischief_count

Number of mischief offenses

rate_per_100k

Mischief rate per 100,000 population

pop_proportion

Proportion of total Canadian population

mischief_proportion

Proportion of total mischief offenses

Source

Correll & Heer (2017). Surprise! Bayesian Weighting for De-Biasing Thematic Maps. IEEE InfoVis.

Examples

data(canada_mischief)

# Basic exploration
head(canada_mischief)
#>               name population mischief_count rate_per_100k pop_proportion
#> 1          Ontario   13448494          45123      335.5246     0.38235126
#> 2           Quebec    8164361          28456      348.5392     0.23211920
#> 3 British Columbia    4631302          22145      478.1593     0.13167156
#> 4          Alberta    4067175          18234      448.3210     0.11563298
#> 5         Manitoba    1278365           9876      772.5493     0.03634492
#> 6     Saskatchewan    1098352           8123      739.5625     0.03122701
#>   mischief_proportion
#> 1          0.31378344
#> 2          0.19788182
#> 3          0.15399540
#> 4          0.12679847
#> 5          0.06867729
#> 6          0.05648700

# Compute surprise
result <- auto_surprise(
  observed = canada_mischief$mischief_count,
  expected = canada_mischief$population
)

# See which provinces are most surprising under the selected models
canada_mischief$surprise <- result$surprise
canada_mischief$signed_surprise <- result$signed_surprise
canada_mischief[order(-abs(canada_mischief$signed_surprise)), ]
#>                         name population mischief_count rate_per_100k
#> 1                    Ontario   13448494          45123      335.5246
#> 2                     Quebec    8164361          28456      348.5392
#> 4                    Alberta    4067175          18234      448.3210
#> 3           British Columbia    4631302          22145      478.1593
#> 13                   Nunavut      35944            678     1886.2675
#> 7                Nova Scotia     942926           4321      458.2544
#> 8              New Brunswick     753914           3456      458.4077
#> 5                   Manitoba    1278365           9876      772.5493
#> 6               Saskatchewan    1098352           8123      739.5625
#> 11     Northwest Territories      43283            456     1053.5314
#> 12                     Yukon      35874            234      652.2830
#> 10      Prince Edward Island     146447            567      387.1708
#> 9  Newfoundland and Labrador     526702           2134      405.1627
#>    pop_proportion mischief_proportion  surprise signed_surprise
#> 1     0.382351260         0.313783440 0.6176481      -0.6176481
#> 2     0.232119203         0.197881824 0.6018849      -0.6018849
#> 4     0.115632983         0.126798467 0.5882624       0.5882624
#> 3     0.131671558         0.153995396 0.5876511       0.5876511
#> 13    0.001021916         0.004714783 0.5860292       0.5860292
#> 7     0.026808128         0.030048052 0.5856093       0.5856093
#> 8     0.021434368         0.024032878 0.5854757       0.5854757
#> 5     0.036344922         0.068677288 0.5853355       0.5853355
#> 6     0.031227011         0.056487000 0.5851333       0.5851333
#> 11    0.001230570         0.003171005 0.5850940       0.5850940
#> 12    0.001019926         0.001627226 0.5849625       0.5849625
#> 10    0.004163603         0.003942894 0.5543334      -0.5543334
#> 9     0.014974552         0.014839746 0.5515620      -0.5515620