Abstract: In view of problems of low visualization degree,low outburst prediction accuracy and low real-time performance in existing coal and outburst prediction methods,a multivariate data visualization system for coal and gas outburst prediction is constructed taking Xinyuan Coal Mine as a test mine.The system obtains basic data involved in outburst prediction through the ways of geophysical exploration, drilling, etc. and WTC gas outburst parameter meter, safety monitoring system, and special data acquisition instrument, and transmits the outburst prediction data through underground industrial ring network and ground office network; two-level prediction method of regional overall control and local online identification covering the spatio-temporal evolution relationship of outburst danger is adopted,so that the messy original data becomes orderly through deep mining of multivariate data of coal and gas outburst,laying the foundation for data visualization; the data of regional prediction subsystem and local prediction subsystem is effectively fused through integrated visualization platform of outburst prediction based on WebGIS to realize intuitive display of prediction results and multivariate data, so that the outburst prediction process can be controlled and the results can be checked. The application results show that the system changes the single and discontinuous status of outburst prediction index of Xinyuan Coal Mine, and significantly improves the accuracy and real-time performance of coal and gas outburst prediction.