handle_missing.py 文件源码

def knn(data, predict=False, best_n=None):
    if best_n:
        # prediction
        clf = KNeighborsClassifier(n_neighbors=best_n)
        return clf
    knn_scores = []
    for n_neighbors in range(4, 51):
        clf = KNeighborsClassifier(n_neighbors=n_neighbors)
        scores = cross_val_score(clf, data.X_train, data.y_train, cv=5)
        knn_scores.append((n_neighbors, scores.mean()))
    knn_scores = sorted(knn_scores, key=lambda x: x[1], reverse=True)
    print(knn_scores)