The continuous development and application of the Internet of Things and blockchain technology have greatly improved the quality and efficiency of demand analysis, and increased the success rate of projects. However, the complexity of financial decision support systems is higher, and the concealment and variability of requirements further increase the difficulty of such systems. The database server and IoT (Internet of Things) server in this article were directly connected to the switches of provincial branches, while the business decision-making terminals of provincial branches were connected to the switches of provincial branches through routers and firewalls. This architecture clearly and accurately reflects the decision-making levels of provincial branches at various levels of business decision-making. In the comparison of results between multiple random decision trees, C4.5 algorithm, and stratified random sampling method, stratified random sampling had the highest classification accuracy, with a low risk of stratified random sampling and a highest classification accuracy of 97.8%.