人工神经网络和随机森林在回归问题中的应用比较

Comparison of Artificial Neural Network and Random Forest in Regression Problems

机器学习方法在回归问题中的应用十分广泛,人工神经网络(Artificial Neural Network,ANN)和随机森林(random forest,RF)均是经典的机器学习算法,在回归问题中均有众多的应用。神经网络和RF算法均为决策树算法的扩展,且均在解决回归问题中有着良好的精度。ANN是一种可以广泛应用于各个学科的经典机器学习算法;RF算法具有结构清晰、易于解释、运行效率高且对于数据要求低等优势,且RF模型具有稳定性较高,不易出现过拟合问题等特点。文章通过2个回归问题的案例,比较神经网络和RF算法在回归问题中的区别,为研究2种算法在回归问题中的应用提供参考。

The machine learning method is widely used in regression. Artificial neural network(ANN) and random forest(RF) are classical machine learning algorithms widely applied in regression problems. Both neural network and RF algorithm are extensions of decision tree algorithm, and both of them have good accuracy in solving regression problems. ANN is a classical machine learning algorithm which can be widely used in various disciplines, RF algorithm has the advantages of clear structure, easy interpretation,high running efficiency and low data requirements, and the RF model has high stability. It is not easy to have the characteristics of over-fitting problem and so on. In this paper, two cases of regression problems are used to compare the difference between neural network and RF algorithm in regression problems, which provides a reference for the study of the application of the two algorithms in regression problems.