Reconcile 990 rows to accounting identities with the least change
reconcile.RdAdjusts values as little as possible (weighted least squares) so that each
row satisfies the bundled accounting_identities. Columns named in fixed
are held exactly; among the rest, weights set how resistant each variable is
to change (larger = moves less). Only rows with no missing values in the
relevant fields are reconciled.
Usage
reconcile(
data,
section = NULL,
fixed = NULL,
weights = NULL,
rows = NULL,
id = "EIN2",
time = "TAX_YEAR"
)Arguments
- data
A data frame of 990 financial fields.
- section
Identity sections to enforce.
NULL(default) uses all.- fixed
Character vector of columns to hold fixed (e.g. reported totals).
- weights
Optional named vector of per-variable weights (default equal).
- rows
Rows to reconcile: a logical/integer index. Default is rows where
imputed_rowisTRUEif that column exists, otherwise all rows.- id, time
Identifier columns (unused by the math; kept for symmetry).
Value
data with reconciled values; an "reconciled" attribute records how
many rows were adjusted and how many were skipped for missing values.