天津市物流需求影响因素的分析和预测研究
Forecast Analysis of Logistics Demand in Tianjin Based on Neural Network

本文以物流需求理论为基础，以天津市的物流需求为例，运用Matlab软件和SPSS软件进行分析，建立物流需求预测模型。首先运用SPSS软件分析了天津市1978~2019年的物流需求指数的线性相关性，结果表明，天津市货物运输量与各经济指标之间存在极强的相关性，说明了选取指标的有效性。其次，本文从天津市物流需求影响因素多元化角度出发，建立了多元回归模型对物流需求影响因素进一步分析，对归一化后1978~2014年的数据进行训练，2015~2019年的数据进行预测，得出对物流需求影响最大的三项经济指标为：外贸进出口总额、第三产业产值、CPI。然后引用trainlm函数建立BP神经网络模型，结合先前的多元回归预测模型得到2种预测结果，与实际值相比较，检验模型的有效性。最后，根据本文研究所得到的结果，综合众多因素为当地政府在物流需求领域提出实时建议。
Based on the theory of logistics demand, this paper takes the logistics demand of Tianjin as an ex-ample, analyzes the Matlab software and SPSS software, and establishes the logistics demand pre-diction model. Firstly, the linear correlation of the logistics demand index of Tianjin from 1978 to 2019 was analyzed by SPSS software, and the results showed that there was a strong correlation between the cargo transportation volume and various economic indicators in Tianjin, which ex-plained the effectiveness of the selected indicators. Secondly, from the perspective of the diversi-fication of the influencing factors of logistics demand in Tianjin, this paper establishes a mul-ti-regression model to further analyze the influencing factors of logistics demand, trains the data from 1978 to 2014 after normalization, and forecasts the data from 2015 to 2019, and draws that the three economic indicators with the greatest impact on logistics demand are: total foreign trade import and export, tertiary industry output value, and CPI, then reference the trainlm function to establish the BP neural network model, combine the previous multiple regression prediction mod-el to obtain two kinds of prediction results, compare with the actual value, and test the effective-ness of the model. Finally, based on the results of the research in this paper, a combination of fac-tors provides real-time recommendations for local governments in the field of logistics needs.