The purpose of this article is to introduce the design and implementation for Securities Company Customer Analysis System based on data mining technology, which focuses on system principles, design ideas, the securities characteristics data mining model, its application, and other important content. With the k-Means clustering method to build a customer preference segmentation model, the customers are effectively divided into eight groups. Using the Decision Tree Combining with the Logistic regression method to Construct the Customer Loss early-warning model is constructed. The results show that the rate of Customer Loss Capture significantly increases after implementing the customer loss early-warning model.