Research Of Railway Freight Volume Forcasting Based On Data Mining

Railway freight volume as an important index in railway freight transportation market system, it largely reflects the national economy’s important demand for railway transport of goods, and it is the important basis of arranging goods transportation plan, so the railway internal departments should closely cooperate with each other. Under the condition of market economy, railway departments need to analyze railway transport demand forecast timely, accurately and scientifically. The trend of railway freight volume is the outcome of combined action of internal and external factors, the mapping relationship between each other is closely, so we need to clarify the complex relation, and make accurate judgment for railway freight volume, thus to promote the stable development of railway freight.This paper is based on analysis of data mining technology and common estimate method, to set freight volume forecast goal, and design data warehouse according to goal. The Multi-Dimensional Data Set displays visualization of Guangzhou-Shenzhen line freight volume, we can observation freight volume change characteristics of stations and railway line, and then, determine the analysis themes are monthly forecast and annual forecast, and respectively adopt different mathematical model. To monthly forecast of freight volume, we choose holt-winters model and SARIMA model, considering the seasonality, trend and stability factor on the changes of freight, to calculate specific monthly forecast results; To annual forecast of freight volume, we use the time series model and Neural network model, to respectively analyze the single factor and multiple factor’s influence on railway freight volume, and compare with the prediction result. This article also analyses the prediction results, and puts forward the applicable for a forecast target scene.