Add an aggregation rule
add_trait_aggregation.RdAdd an aggregation rule
Usage
add_trait_aggregation(
source_trait_id,
method,
target_trait_id = NULL,
method_param = NULL,
min_n = 1L,
notes = NULL,
con = NULL
)Arguments
- source_trait_id
Integer. Trait id whose individual-level rows (in `data_traits_measures.traitid`) feed the aggregate.
- method
Character. One of the supported methods (see `.compute_aggregate()`).
- target_trait_id
Integer or NULL. Trait id stamped on the aggregated output rows in `taxa_traits_measures.fk_id_trait`. Three usage modes:
* **`NULL` (default) — auto-derive.** Creates (or reuses) a derived `traitlist` entry named `<source_trait>_<suffix>` (e.g. `stem_diameter_p95`, `stem_diameter_max`) with the source's `valuetype`, `expectedunit`, and value bounds copied over. The derived trait's id is stored as `target_trait_id`. The method is preserved in the trait name and multiple aggregations of the same source coexist as distinct columns in `query_taxa_traits()`. * **`source_trait_id` — keep the original trait name.** The aggregated row is written under the source trait id, so wide pivots show it as `taxa_<source_trait>` and there is no name suffix. Trade-offs: (a) the aggregation method is no longer visible at the trait-id level — it survives only in `measurementremarks` (e.g. `"percentile(n=12) param=95"`) and `basisofrecord = "AggregatedFromIndividual"`; (b) you can only have one such rule per source trait — two rules pointing to the same id collide in `(idtax, fk_id_trait)` and the wide pivot averages them; (c) any directly-measured taxa-level rows for that trait are mixed in. * **An explicit integer** — use that trait id (e.g. a pre-existing custom entry in `traitlist`) without auto-creation.
- method_param
Numeric or NULL. Required for `"percentile"`.
- min_n
Integer. Minimum non-NA values per taxon to emit a row (default 1).
- notes
Character. Free-text description.
- con
Connection (admin/write privileges).