Result of a causal effect query — EffectResult • kagu

Returned by KaguModel$effects(). Holds the full posterior over the causal effect along with metadata.

Scalar effect: samples is a [n_chains, n_draws] matrix.
Sweep effect: samples is a [n_sweep, n_chains, n_draws] array.

Public fields

source: Character - the intervention node.
target: Character - the outcome node.
from_value: Numeric - intervention baseline value (or sweep grid).
to_value: Numeric - intervention target value (or sweep grid).
samples: Numeric array - posterior effect samples.
hdi_prob: Numeric - HDI probability used for summaries and plots.
std_units: Logical - whether the effect is in SD units.
conditions: Named list of conditioned node values, or NULL.
sweep_values: Numeric vector of sweep grid values, or NULL.
conditions_label: String representation of conditions (for plotting).
compare_results: List of other EffectResults (for multi-condition plots).

Methods

`EffectResult$new()`

Create an EffectResult (normally called by compute_effect).

Usage

EffectResult$new(
  source,
  target,
  from_value,
  to_value,
  samples,
  hdi = 0.9,
  std_units = FALSE,
  conditions = NULL,
  sweep_values = NULL
)

Arguments

source: Character - the intervention node.
target: Character - the outcome node.
from_value: Numeric - intervention baseline value.
to_value: Numeric - intervention target value.
samples: Numeric array - posterior effect samples.
hdi: Numeric - HDI probability.
std_units: Logical - whether the effect is in SD units.
conditions: Named list of conditioned node values.
sweep_values: Numeric vector of sweep grid values.

`EffectResult$is_sweep()`

Is this a sweep result?

Usage

EffectResult$is_sweep()

Returns

Logical scalar.

`EffectResult$summary()`

Tabular summary of the effect posterior.

Returns a tibble, consistent with KaguModel$summary(). The HDI columns use the same hdi_lower / hdi_upper naming as build_summary_table().

Scalar: a one-row tibble with source, target, from, to, mean, sd, hdi_lower, hdi_upper.
Sweep: one row per grid point with source, target, x, mean, sd, hdi_lower, hdi_upper.

Usage

EffectResult$summary()

Returns

A tibble.

`EffectResult$plot()`

Plot the posterior effect distribution.

Scalar: density plot with mean point and HDI bar.
Sweep: dose-response line with HDI ribbon.

Usage

EffectResult$plot(compare = NULL, base_label = NULL)

Arguments

compare: Optional named list of other EffectResult instances to overlay (useful for comparing sweeps under different interactions/conditions). Only valid if this result and all comparison results are sweep = TRUE. If this effect was generated with a list of conditions, compare is populated automatically.
base_label: Optional character string to label this EffectResult in the legend when plotting with compare (defaults to "Base" or the condition label).

Returns

A ggplot object.

`EffectResult$diagnostics()`

R-hat and ESS diagnostics for the effect posterior.

Chain structure is preserved, so r-hat values are genuine.

Usage

EffectResult$diagnostics()

Returns

A tibble from posterior::summarise_draws().

`EffectResult$print()`

Print method - shows the summary table.

Print summary of the effect result.

Usage

EffectResult$print(...)

Arguments

...: Ignored.

`EffectResult$clone()`

The objects of this class are cloneable with this method.

Usage

EffectResult$clone(deep = FALSE)

Arguments

deep: Whether to make a deep clone.