Pettitt-style change-point on yearly incident counts — morie_tps_changepoint

Implements Pettitt's non-parametric change-point statistic U_t = sum_i sum_j sign(x_i - x_j) for i <= t < j, then reports the year maximising |U_t| and an approximate p-value. No external change-point dependency is required.

Usage

morie_tps_changepoint_detection(df, year_col = "OCC_YEAR", ds_name = "?")

Arguments

df: A data.frame with one row per incident.
year_col: Year column name.
ds_name: Character label.

Value

A morie_rich_result list with changepoint_year, K_statistic, p_value, pre_mean, post_mean.

Examples

set.seed(1)
df <- data.frame(OCC_YEAR = rep(2014:2023, each = 30),
                 OCC_MONTH = sample(month.name, 300, TRUE),
                 HOOD_158 = sample(sprintf("%03d", 1:20), 300, TRUE),
                 LAT_WGS84 = runif(300, 43.6, 43.8),
                 LONG_WGS84 = runif(300, -79.5, -79.2))
res <- try(morie_tps_changepoint_detection(df, ds_name = "synthetic"))
if (!inherits(res, "try-error")) str(res, max.level = 1)
#> List of 13
#>  $ title           : chr "Change-point (Pettitt) -- synthetic"
#>  $ call            : chr "morie_tps_changepoint_detection(df=<300r>, year_col=OCC_YEAR)"
#>  $ summary_lines   :List of 8
#>  $ warnings        : chr(0) 
#>  $ interpretation  : chr "Estimated structural break in 2014: pre-mean 30.0, post-mean 30.0. p=1.0000 -- not significant at alpha=0.05."
#>  $ n               : int 10
#>  $ years           : int [1:10] 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023
#>  $ counts          : int [1:10] 30 30 30 30 30 30 30 30 30 30
#>  $ changepoint_year: int 2014
#>  $ K_statistic     : num 0
#>  $ p_value         : num 1
#>  $ pre_mean        : num 30
#>  $ post_mean       : num 30
#>  - attr(*, "class")= chr [1:3] "morie_tps_temporal_result" "morie_rich_result" "list"