经济预测是人类依据对经济活动运行规律的认知,对经济运行相关领域未来发展方向与所处位势的可能性进行推断.经济预测对运行规律的尊重表现为需要从数据和历史事实出发,应用合乎逻辑的方法构建经济预测模型.由于经济活动运行规律时间不变性较短,而且运行规律本身就包括着随机性,经济预测不是一个确定性的预测,而是一种可能性的预测.因经济活动主体运行规律各有不同,经济预测方法丰富多样.文章从预测方法的技术属性角度,按照时间序列方法、结构方程方法、信号特征方法、机器学习方法和组合预测方法等几种分类对经济预测主要定量方法的原理进行了系统的梳理,简单评价了各类定量预测方法的适用范围,并对经济预测定量方法进一步发展进行了展望.

Economic forecasting is to make judgement about the possibility of the future trend of economic development, based on the human beings' knowledge about economic activities. It is crucial for economic forecasting to build economic forecasting models based on data and historical facts under guidance of economic logics and economic laws. Because of the short invariance of economic laws and the randomness embedded into economic laws, economic forecasting is not a deterministic prediction, but a possible prediction. Different economic activities may have different logics, the economic forecasting methods are rich but have their specific application areas. In this paper, we divide the forecasting methods into the following classes: Time series methods, structural equation methods, signal methods, machine learning methods and combination forecasting methods. We give a brief introduction of each method, and discuss the possible application area and pros and cons of each class. We also give an outlook about the future development of forecasting methods.