应用传统统计技术常会因样本小和测量数据不符某种分布而受到限制．本文评价一种前馈型神经网络算法以预测落叶阔叶林产量．另外，还介绍一种由定性变为定量的数据变换方法，以用相对小的样本建立多元回归预测模型．数据变换方法有助于改善多元回归模型的预测效果．在本实验的条件下，研究结果表明神经网络技术能够产生最好的预测效果 Conventional statistical techniques are often limited by the fact that the sample size is small and the measurement data does not match a certain distribution. This paper evaluates a feedforward neural network algorithm to predict deciduous broad-leaved forest yield. In addition, a method of data transformation from qualitative to quantitative is introduced to establish multiple regression prediction models with relatively small samples. Data transformation methods can help improve the prediction of multiple regression models. Under the conditions of this experiment, the results show that neural network technology can produce the best prediction effect