多因素模糊时间序列预测模型的研究

发布时间：2019-04-12 09:59

【摘要】：模糊时间序列因为在解决模糊性数据和不确定性数据方面的优势,以至于很多研究人员建立模糊时间序列模型处理预测问题。模糊时间序列模型在各个领域都有广泛应用,提高预测精度一直是研究者关注的焦点。由于传统的模糊时间序列模型大都是处理单因素问题的模型,而在现实生活中几乎没有一个现象是独立发生的,与其他因素都有着或多或少的联系。建立多因素模糊时间序列模型成为了亟待解决的问题,在模糊时间序列模型中,提高模型预测精度的重要部分是模糊关系的建立和应用、论域划分、数据模糊化。故本文建立多因素模型的主要工作围绕在论域划分和模糊关系建立和应用两个方面。在论域划分方面,在经典的模糊时间序列模型中通常使用划分论域构建模糊集,通过选择适当的间隔数和论域长度提高预测的精确性。基于此,本文提出基于空间FCM的多因素模糊时间序列模型,通过空间距离更好的反映了多因素数据在空间的分布,真实的反映了数据结构,体现了数据的敏感性。最后利用亚马逊公司股指的收盘价、最高价和最低价对收盘价进行预测,验证了该算法的有效性。在模糊关系建立和应用方面,本文在分析经典模糊时间序列预测模型的基础上,结合证据理论提出基于均分论域和证据理论的多因素模糊时间序列模型,基于空间FCM和证据理论的多因素模糊时间序列模型,基于可调参的证据理论的多因素模糊时间序列模型,最后通过亚马逊公司股指的收盘价的预测结果验证了本文模型的理论可行性和有效性。
[Abstract]:Because of the advantages of fuzzy time series in solving fuzzy data and uncertain data, many researchers set up fuzzy time series models to deal with prediction problems. Fuzzy time series models are widely used in various fields, and improving prediction accuracy has always been the focus of researchers. Since most of the traditional fuzzy time series models deal with single-factor problems, almost none of the phenomena occur independently in real life, and they are more or less related to other factors. The establishment of multi-factor fuzzy time series model has become an urgent problem. In the fuzzy time series model, the establishment and application of fuzzy relations, domain division and data fuzzification are the most important parts to improve the prediction accuracy of fuzzy time series model. Therefore, the main work of establishing multi-factor model in this paper is divided into two aspects: domain division and fuzzy relation establishment and application. In the aspect of domain partition, the partition universe is usually used to construct fuzzy sets in classical fuzzy time series models, and the accuracy of prediction is improved by selecting the appropriate interval number and the length of the domain. Based on this, a multi-factor fuzzy time series model based on spatial FCM is proposed in this paper, which reflects the spatial distribution of multi-factor data better through the spatial distance, reflects the data structure and reflects the sensitivity of the data. Finally, the closing price, the highest price and the lowest price of Amazon stock index are used to predict the closing price, and the validity of the algorithm is verified. Based on the analysis of the classical fuzzy time series prediction model and the evidence theory, a multi-factor fuzzy time series model based on the equally divided domain and the evidence theory is proposed in this paper, which is based on the establishment and application of fuzzy relations. Multi-factor fuzzy time series model based on spatial FCM and evidence theory, and multi-factor fuzzy time series model based on adjustable parameter evidence theory. Finally, the theoretical feasibility and effectiveness of the model are verified by the prediction results of Amazon stock index closing price.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：O159;O211.61

【相似文献】