Research On Application Of Key Technologies Based On Multi-source Data Association Analysis

Association analysis is an important technique for discovering interesting associations between large amounts of data.At present,it has been applied to the fields of commerce,telecommunications,finance,agriculture,and medical treatment,and has achieved good results.The current increase in the number of motor vehicles has led to frequent occurrences of various types of traffic accidents.Correlation analysis of existing accidents and accident prediction based on association rules can provide relevant departments with information to avoid traffic accidents to a greater extent.The purpose of this work is to study a more efficient association rule mining algorithm,to establish an association analysis prediction model that can ensure its accuracy even under unbalanced data sets,and to apply the model to traffic accident prediction.The main work of the thesis includes the following points:(1)An improved association classification rule mining algorithm is proposed.In this thesis,an improved equivalence-type rule tree structure is proposed for mining association rules.Based on the equivalence-type rule tree,prior rules and diffset strategies are used to implement deep pruning of rules.thereby reducing the running time of the algorithm.(2)An association analysis prediction model based on multi-source data is proposed.This article uses a combination of multiple metrics and a multi-criteria decision algorithm(ELECTRE TRI)to perform rule screening,proposes an improved Laplace metric for rule ranking,and uses a database overlay method for data coverage to obtain strong association rules to build a kind of Association analysis prediction model applied to unbalanced data.(3)Apply the model to actual traffic accident prediction and design and implement a traffic accident prediction system to visualize the prediction results.This thesis designs and implements a traffic accident prediction system based on the above association rule mining algorithm and association analysis and prediction model,and realizes the real-time prediction function of possible traffic accidents.The overall system design and system function modules are described in detail.The actual effect was demonstrated.The algorithms and models proposed in this thesis have verified their effectiveness through the analysis of their respective results,and applied them to real traffic accident prediction scenarios,indicating that the proposed algorithms and models can play their role in real application scenarios and have certain effects.