基于出租车GPS地理位置的打车推荐算法 Zhi-Qiang You & Lu Yu 2013/11/16 移动互联的数字化生活 打车推荐算法框架 GPS与数字地图匹配 近几年的相关研究及展望 手机 谷歌眼镜 智能手表 全能型穿戴设备 Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Taxi Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Taxi Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Taxi Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Taxi Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Taxi Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Taxi Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Taxi Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Passenger Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Passenger Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Passenger Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Passenger Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Passenger Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Online Passenger Recommendation Jing Yuan*, Yu Zheng, Liuhang Zhang, Xing Xie, Guangzhong Sun, Where to Find My Next Passenger? , 13th ACM International Conference on Ubiquitous Computing (UbiComp 2011), Beijing, China, Sep. 2011. Map-Matching (GPS地图匹配算法) Map-matching is the process of aligning a sequence of observed user positions with road network on a digital map. Moving Object management Fundamental pre-processing Step for Traffic flow analysis Driving directions Yin Lou et al. Map-Matching for Low-Sampling-Rate GPS Trajectories. (ACM SIGSPATIAL GIS 2009). Yuan J, Zheng Y, Zhang C, et al. An interactive-voting based map matching algorithm[C]//Mobile Data Management, 2010 Eleventh International Conference on. IEEE, 2010: 43-52. Two Situations 1.High Sampling Points(typically one point every 10-30s) The incremental algorithm Traditional Algorithms Average-Frechet-Distance based global map-matching Algorithm Work Well ! 2.Low Sampling Points(one point every 2-5 minutes) Uncertainty goes up With such sampling rate,the distance between two points may reach over 1300m even a vehicle’s speed is only 40km/h Incremental and AFD algos Perform bad ! Under such circumstance , Yin propose a novel global map-matching algorithm called ST-Matching for low-sampling-rate GPS trajectories! Smile again! BTW 1.Local / imcremental methods The approaches which address the map-matching problems can be generally classified into 2. global methods three classes 3.Statistical models Now , you must be clear: ST-Matching algorithm 1. be created to address the low-sampling-rate GPS Trajectory matching problem. 2. belongs to the global algorithm class. 3. aim to improve the accuracy of map-matching under low-sampling-rate GPS situation. DataSets Taxi GPS Points Road network Now the problem of map-matching is defined as: Given a raw GPS trajectory T and a road network G(V,E),find the path P from G that matches T with its real path. LET’S GO INTO DETAILS Two Observation Of ST-Matching Algorthm SPATIAL ANALYSIS TEMPORAL ANALYSIS Algorithm Architecture Candidate Preparation * 经度:相差3.234059999999772秒,距离相差100.00386455484937 * 纬度:相差7.2766800000096055秒,距离相差100.00159837312073 Candidates num K = 5 If no candidates, return “null”, then ignore this trajectory point. Spatial Analysis Geometric and topological Information of the road network 1.Geometric information is incorporated using observation probability 2.Topological information is expressed using transmission probability 候选点之间的最短路径使用ArcGIS进行求值。 使用的是python接口 Temporal Analysis(but I feel weird about the way they took) Warning: Suppose: Highway: [100,100,100] Service Road:[20,20,20] average speed of taxi: [19,19,19] Similarity results are the same! Inseparable! Result Matching 最长路径算法:基于Dijkstra算法的改型。 Example Evaluation Approaches Obviously, ST-Matching algorithm does’t take other points except the closest points. But: 1.The position context of a GPS point as well as the topological information of road networks 2.The mutual influence between GPS points(the matching result of a point references the positions of its neighbors ; in turn ,when matching its neighbors , the position of this point will also be referenced) 3.The strength of the mutual influence weighted by the distance between GPS points(the farther distance is the weaker influence exists) So , they work out a new algorithm called IVMM(Interactive-voting based map matching algorithm) From the above figure , we can see the first two phases are the same with ST-Matching . Using Spatial and Temporal Analysis , we get the Static Score Matrix. Mutual Influence Modeling Interactive Voting Evaluation Approach 近年来的相关研究 Microsoft Research Asia & Geolife Detect Gathering Patterns Kai Zheng et al. On Discovery of Gathering Patterns from Trajectories, ICDE 2013 Taxi Collaborative Recommendation Nicholas Jing Yuan et al. T-Finder: A Recommender System for Finding Passengers and Vacant Taxis. TKDE 2012 Urban Computing Yu Zheng et al. Urban Computing with Taxicabs, Ubiquitous Computing (Ubicomp) 2011 Map-Matching Jing Yuan et al. An Interactive Voting-based Map Matching Algorithm, IEEE Mobile Data Management (MDM) 2010. ……… Map-Matching (GPS地图匹配算法) ArcGIS @Ersi 地理数据处理软件(ArcGIS) 路网数据格式:1、gdb格式 2、shapefile格式 求解最短路径:提供network analysis和spatial analysis功能 数据导出:XML格式,因此处理路网数据时,格式化文件,那么需要进行超大XML文件解析,我们这次接触的是解析400M左右 的单个XML文件,格式非标准xml,单行显示。所以需要流读取进行处理。Java和python都提供了SAX解析包。 GeoCoding编码以及逆编码 通过地址查询GPS数据 http://maps.googleapis.com/maps/api/geocode/json?address=1600+Amphitheatre+Parkway,+Mountain+View,+CA&sensor= true_or_false 通过GPS数据查询地址(城市、街道等信息) http://maps.googleapis.com/maps/api/geocode/json?latlng=40.714224,-73.961452&sensor=true_or_false 最短路径算法(本算法中,需要想反,使用最长路径算法) 1.Dijkstra's Algorithm 2.A*算法 谢谢!