# train_model

models.train_model(model, df: pd.DataFrame, outcome, features=None, data_type=None, metrics=None, return_preds=False)

This function trains a given model using train/val data.

If outcome is multiple and the model does not support multiple output then an ensemble of models is used.

# Arguments

model: A Model instance. Optionally can be a String to save an import statement, if using default parameters.
df: Pandas DataFrame containing patients and the features to utilize.
features: A list of the feature columns, only if a subset of the df should be utilized. By default all columns within df will be used except for PatientID.
outcome:
- String: If the outcome is a column in the df then provide the name of the column as a String
- List: List of outcomes as strings.
- Pandas Series: For a single output scenario.
- Pandas DataFrame: For multiple output scenarios, each column should be an output.
data_type: An optional Pandas DataFrame containing a column named Type with the values train, val, or test, indicating which dataset the patient belongs in. To use this feature, the outcome cannot be a Pandas Series due to inability to correctly reference patients.
metrics: A list of metrics for the model to be evaluated against and to be printed out. Optionally can be a String to save an import statement, if using default parameters.
return_preds: Whether to return the model predictions. If enabled then returns a tuple containing the train and val predictions.