尽管短期焦炭价格预测难度较大, 但焦炭市场的两项特点很鲜明。第一是焦炭产品的市场化程度很高。焦炭产品价格的形成主要受供需关系影响, 因此对短期焦炭价格规律的研究可以从市场数据的统计模型入手。文献[3-6]的研究结果说明: 半参数模型由于其适用条件宽泛, 且估计效果较好, 常用于各类产品市场的短期价格预测, 所以本文考虑用半参数回归法预测短期焦炭价格。但半参数模型存在边界估计效应且缺乏经济解释意义, 因此如何克服这些缺陷给焦炭短期价格预测带来的影响, 值得本文探讨。深圳英语翻译

It is difficult to predict the short-term price of coke; but the coke markets have 2 striking characteristics. The first characteristic is that coke products are highly marketized. The price of coke product is influenced largely by the supply-demand relation. So, the pattern of the short-term coke price can be researched on the basis of the market data model. The research literature [3-6] reveals that the semi-parametric model is widely applicable, has good prediction results, and is used for prediction of short-term prices of various products on the market. So, in this article semi-parametric regress method is used for prediction of the short-term price of coke. But the semi-parametric regress model can produce boundary estimation errors and is inadequate in economic interpretation role. How to overcome these problems is also worthy of approaching in this article.      

第二是焦炭消费的产业集中性。焦炭作为钢铁冶金行业的主要原材料, 其对焦炭消费比重已达到85%以上^【7】。所以焦炭价格和钢铁价格很有可能存在协整关系。如果两者的协整关系成立, 那么就可以基于协整理论分析, 求出协整系数, 并以这种关系构成误差修正项, 连同其它反映短期波动的解释变量一起, 建立短期预测模型^【8】。

The second characteristic is that the coke is consumed mainly in certain industries. Coke is the main material in the iron and steel making industry. More than 85%^【7】 of coke is consumed in iron and steel making industry. So a co-integration relation is very likely between the coke price and steel/iron price. If the co-integration relation exists, then the co-integration analysis can help produce a co-integration coefficient, and thus to constitute an error correction item, and, together with other variables reflecting short-term fluctuation, to set up a short-term prediction model^【8】.       

综合上述理论分析, 可以考虑把误差修正算子和半参数模型结合, 发挥各自的优点。那么这种改进的新模型对于短期焦炭价格的预测应该有很好的统计性质。至于这种模型的实际应用效果如何, 还得通过实证分析来检验。 广州英语翻译

Based on the above theoretical analysis, consideration can be given to combining error correction operator and semi-parametric model in order to give play to each other's advantages. Then, this improved model is statistically good for prediction of short-term coke price. As for the actual effect of the model, example-based analysis therefor is necessary.