To improve emergency evacuation capacity the objectivity of the evaluation index system and evaluation method, this article constructed an evaluation index system of subway station emergency evacuation capacity from four aspects: management factors, emergency response, building facilities and personnel evacuation. In view of the passenger flow characteristics and passengers' unsafe behaviors, the variable fuzzy set theory was introduced, and the minimum relative information entropy combined with analytic hierarchy process and entropy weight method was used to optimize the combination of subjective and objective weights, which can effectively solve the shortcomings of linear weighting method in the determination process and stability of weighting coefficients. Taking Xi 'an Road subway station in Dalian as an example, the emergency evacuation capacity of Xi 'an Road subway station was evaluated, and the emergency evacuation capacity of Xi 'an Road subway station was obtained to be Ⅱ level (good), and it had a trend of development to be Ⅰ level (great), which verified the scientific rationality of the model.

Data mining refers to the process of searching the massive data stored in the data warehouse through related algorithms and obtaining the hidden information.Cluster analysis is one of the main technologies in data mining, which is widely used in the field of financial investment.This paper takes similarity as evaluation standard, uses cluster analysis method to detect and analyze various stock indicators, and provides corresponding investment suggestions according to the actual situation.Experiments show that this method can help users to make reasonable prediction of the growth trend of stocks, and finally assist them to make the right investment decision.The experimental results show that this method is more convenient and effective than other methods

In recent years, with the rapid development of technology, location-based services have brought many conveniences to people's lives, but also brought risks of privacy leakage. Attackers can use users' trajectory information to deduce their home address, work address and other private information, which poses significant hazards and risks to users' lives. To address this challenge, there is a personalized privacy protection scheme that marks the semantic attributes of users' trajectory points based on existing POI semantic information. The resulting set of semantic attributes is classified and abstracted to obtain a classification tree, and three types of stop points are designed to select appropriate regions of POIs for replacement to generate fake trajectories. Finally, the fake trajectories are tested for obstacle crossings, and if they cross any obstacles, POIs are reselected for replacement. This article analyzes and simulates this scheme.