Abstract: As a solution to the massive marketing data and unsatisfactory analysis efficiency, a visual analysis system of marketing data (VASMD) for cigarette enterprises was developed based on thermodynamic diagram. VASMD adopted a triple-layer distributed data processing architecture, including data collecting layer, data preprocessing layer, and visual analysis layer. Data analysis models were built by using k-means and DBSCAN data mining algorithms, an interactive visual analysis diagram and several auxiliary analysis tools were developed based on Baidu map. In the case studies, two marketing data sets of cigarette brand "Zhenlong" produced by China Tobacco Guangxi Industrial Limited Corporation were used for screening, preprocessing and visual analysis. The results showed that: 1) The development of sale hot spots in main cities had strong guiding effects on the hot spot formation of their neighboring counties, towns and villages. 2) The hot spots distributed and spread mostly along the major and trunk traffic networks. 3) There existed mutual traction effects between those regions with frequent economic contacts and similar language and culture in the formation and spreading of hot spots. In conclusion, a VASMD in the form of thermodynamic diagram can figuratively demonstrate the geographic distribution and structure of cigarette consumer groups, reveal the market development patterns and rules implied by the data, and promote the level of precise marketing of enterprises effectively.