ExperimentDesigner.recommend

ExperimentDesigner.recommend(spend_changes=None, durations=None, min_snr=2.0, significance_level=0.05, score_weights=None)

Recommend experiments across all channels.

Evaluates a grid of candidate designs (channel x spend change x duration) and returns a ranked collection of recommendations.

Parameters:

spend_changeslist[float] | None

Fractional per-week spend changes. E.g. [0.2, -0.5, -1.0] means +20%, -50%, and go-dark. Defaults to [0.1, 0.2, 0.3, 0.5, -0.2, -0.5, -1.0].

durationslist[int] | None

Experiment durations in weeks. Defaults to [4, 6, 8, 12].

min_snrfloat

Minimum signal-to-noise ratio to include in results.

significance_levelfloat

Significance level for the two-sided power calculation.

score_weightsdict[str, float] | None

Custom weights for scoring dimensions. Keys are "uncertainty", "correlation", "gradient", "assurance", "cost_efficiency".

Returns:

ExperimentRecommendations

Recommendations sorted by score (descending).