mcf_main.ModifiedCausalForest.train_iv

ModifiedCausalForest.train_iv(data_df)

Train the IV modified causal forest on the training data.

Parameters

data_df (DataFrame) – Data used to compute the causal forest. It must contain information about outcomes, treatment, and features.

Returns

results – Contains the results. This dictionary has the following structure: ‘tree_df’ : DataFrame

Dataset used to build the forest.

’fill_y_df’DataFrame

Dataset used to populate the forest with outcomes.

’common_support_probabilities_tree’: pd.DataFrame containing

treatment probabilities for all treatments, the identifier of the observation, and a dummy variable indicating whether the observation is inside or outside the common support. This is for the data used to build the trees. None if _int_with_output is False.

’common_support_probabilities_fill_y’: pd.DataFrame containing

treatment probabilities for all treatments, the identifier of the observation, and a dummy variable indicating whether the observation is inside or outside the common support. This is for the data used to fill the trees with outcome values. None if _int_with_output is False.

’path_output’Pathlib object

Location of directory in which output is saved.

Return type

Dictionary.