Abstract: From the goal of sub-line cost accounting to carry out data analysis, this article built the overall structure of data analysis, described the relationship between data analysis and sub-line cost accounting system, introduced the data mining algorithms such as clustering, classification, regression analysis and correlation analysis, analyzed the characteristics of public welfare / business characteristics of railway lines, the distribution characteristics of different cost types, the distribution rules of various kinds of statistical workload, the profit and loss situation of the railway and so on, the railway company / the railway station / the company and the specific line. It was provided data and support decision support for the rational definition of the nature of the railway line business, transport product optimization, business decision-making.