Design And Implementation Of The Subway Comprehensive Service System Based On Android Platform

The development of science and technology has gradually changed the ways of people to travel,more and more people choose to travel by subway.However,most relevant subway software only provides route planning services for users,and it is increasingly unable to meet users' demands for multi-dimensional subway travel.At the same time,with more and more careful travel planning,the accuracy of subway route recommendation becomes more and more important.Especially for some users who have the need of fine subway routine query,such as those who catch the last subway,those who rush to work in time,etc.However,most of the recommendations software for subway routine on the market ideally estimates the travel time,which can not meet the increasing demands of users for the accuracy of prediction.Based on the above reasons,this paper is committed to building a subway full-service system platform,not only provides users with intelligent path planning function,but also provides users with more refined services related to riding the subway.For the intelligent route planning function in the system,mainly through the processing and analysis of railway station data,and combined with the application of gradient boosting decision tree,k-means clustering and linear regression model in machine learning in the field of passenger flow prediction.This module also employs the Dijstra algorithm,and it mainly provides following functions:subway optimal path based on least transfers,least cost,and latest departure time.In the case of unreachable,this module will provide the subway optimal path to the station that:is the nearest to the destination.In addition,to meet the other requirements of the user travel,the system also provides users with the article reading,the article recommended,the lost and found,surrounding facilities,information display,VR station wizard,delay,current-limiting push function,such as the above functions are greatly improved the users retained and active in the system.The overall structure of the system adopts C/S mode,the article recommendation module is based on lambda architecture,which is divided into offline batch processing layer,real-time recommendation layer and service layer.According to the user's browsing records and interests,recommend relevant articles for users,enrich the user's ride experience.The Android development of other modules follows the design philosophy of MVC and uses Spring Boot as the server-side framework,and MySQL database,HDFS files are used to provide persistent storage of data,and use Redis as a cache.At present,this system have been put on line.The average query time of the subway route recommendation module is 39.3 ms,and the accuracy of the model(the path with error less than three minutes is the exact path)reaches 90.5%.The accuracy of the model is relative high,which basically meets the requirements of the system.