Cross-fitted (out-of-fold) propensity scores from an EconML DR-style estimator. Each row's prediction comes from the fold model whose trainin

Average feature importances across EconML cross-fitted nuisance models. Parameters ---------- models_nested : nested list Ex

Heuristically determine whether a pandas Series is probably continuous. This avoids scanning the full column by using at most `sample_size`

Derived bool: True iff no enabled estimand is `not_satisfactory`.

Construct the covariate matrix used by EconML nuisance models. If both X and W are supplied, EconML nuisance models generally use the co

Convert a mixed-type pandas DataFrame into a fully numeric DataFrame. Parameters ---------- df : pd.DataFrame Input datafram

Partition by a boolean cut and print two-row tables per estimand.

Inject `spec` parameters into the notebook's top configuration cell, append a results serialization cell that writes results.json, execute the

Compute standardized mean differences for binary treatment. SMD = (mean_treated - mean_control) / pooled_sd

Compute weighted standardized mean differences for binary treatment.

Worst tier across enabled estimands — handy for one-line summaries.