Skip to contents

create dummy transaction data

Source: R/create_dummy_data.R
create_dummy_transaction_data.Rd

create dummy transaction data

Usage

create_dummy_transaction_data(
  people = 100,
  size = 2,
  spatiotemporal = FALSE,
  places = 50,
  days = 30,
  seed = NULL
)

Arguments

people

number of people

size

mean record number

spatiotemporal

add the `PLACE` and `TIME` columns described above (default FALSE)

places

number of distinct locations, used only when `spatiotemporal = TRUE` (default 50)

days

length of the observation window in days, used only when `spatiotemporal = TRUE` (default 30)

seed

integer seed, or NULL (the default) to use the ambient RNG stream as the function always has

Value

a tibble with columns ROW_NUMBER, ID, NUM_STATIC, NUM_DYNAMIC, BIN, CHAR – plus PLACE and TIME when `spatiotemporal = TRUE` – containing `people * size` rows of randomly generated dummy transaction data.

Spatio-temporal columns

With `spatiotemporal = TRUE` two further columns are appended, so that [spatiotemporal_unicity()] (Issue #24) has something with the shape of a mobility trace to measure:

`PLACE`

a zero-padded location code, `"P001"` upwards. Codes are assigned so that **sort order tracks proximity** – `"P004"` is next to `"P005"` – because that is what [coarsen_place()] assumes when it merges neighbouring locations into a coarser grid.

`TIME`

hours since the start of the observation window, so a `time_resolution` of 1 means hourly and 24 means daily.

Each person is given a home location, a small repertoire of places around it, and a preferred hour of the day; events are drawn from those habits. Without per-person structure every trace would be a uniform draw from the same pool and the unicity curve would say nothing about the method.

The extra columns are **off by default** so that the returned schema is unchanged for existing callers, and they are drawn *after* the original columns, so a given seed produces the same `NUM_STATIC` / `NUM_DYNAMIC` / `BIN` / `CHAR` either way.

Examples

data_tran = create_dummy_transaction_data(people = 10, size = 4)
create_dummy_transaction_data(people = 10, size = 4, spatiotemporal = TRUE, seed = 1)
#> # A tibble: 40 × 8
#>    ROW_NUMBER    ID NUM_STATIC NUM_DYNAMIC   BIN CHAR  PLACE  TIME
#>         <int> <int>      <dbl>       <dbl> <dbl> <chr> <chr> <dbl>
#>  1          1     9         10      0.245      0 dv    P028    458
#>  2          2     4         10      0.0707     0 bY    P012    690
#>  3          3     7         10      0.0995     0 Wz    P009    217
#>  4          4     1         10      0.316      1 Vg    P046    661
#>  5          5     2         10      0.519      1 bE    P017    181
#>  6          6     7         10      0.662      0 Gj    P009      1
#>  7          7     2         10      0.407      0 SA    P019    665
#>  8          8     3         10      0.913      0 k6    P044    316
#>  9          9     1         10      0.294      1 rc    P001    277
#> 10         10     5         10      0.459      1 YK    P014    691
#> # ℹ 30 more rows