Bias Ratio Between Estimated and Observed Series

Computes mean(estimated) / mean(observed) as a measure of systematic over- or under-prediction. A value of 1.0 indicates no bias. Pairs with calc_model_R2(method = "corr") to give both pattern agreement and magnitude calibration.

Usage

calc_bias_ratio(observed, estimated, na_rm = TRUE, finite_only = TRUE)

Arguments

observed

Numeric vector or matrix.

estimated

Numeric vector or matrix (same length after flattening).

na_rm

Logical; drop NA pairs. If FALSE and any NA is present, returns NA_real_. Default TRUE.

finite_only

Logical; drop non-finite values. Default TRUE.

Value

Scalar ratio. Values > 1 indicate over-prediction, < 1 under-prediction. Returns NA_real_ if observed mean is zero or no valid pairs exist.

See also

calc_model_R2()