main.py 文件源码

def eval_models(eda_objs, clfs):
    '''
    Uses a given set of classifiers objects to evaluates a given set of pipelines
    and return their CV scores.

    Parameters
    ----------
    pipelines_names: list of strings
                names of the pipelines to compare
    eda_objs : list of objects
    clfs     : list of classifiers
    *kwargs : Additional arguments to pass to sikit-learn's cross_val_score 
    '''        

    if isinstance(clfs, list) is False:
        clfs = [clfs]
    acc = []
    for clf_name, clf in clfs:        
        for pipe_name, obj in eda_objs:   
            X, y = obj.df[obj._get_input_features()], obj.df[obj.y]
            cv_score = cross_val_score(estimator=clf, X=X, y=y, cv=5, scoring='r2') #neg_mean_squared_error
            acc.append([(clf_name, pipe_name, v) for v in cv_score])
    acc = [item for sublist in acc for item in sublist] # flatten the list of lists
    return acc