catboost 做回归预测结果出现了负值,在训练集中label中是没有0和负值的,参数如下
model = CatBoostRegressor(
depth=5,
iterations=5000,
learning_rate=0.27,
eval_metric='RMSE',
cat_features=cate_features,
l2_leaf_reg=3,
use_best_model=True,
random_seed=666,
early_stopping_rounds=195,
loss_function='MAPE'
),用xgboost就没有问题,r2分值0.94