DS595/CS525: Urban Network Analysis -- Urban Mobility Prof. Yanhua - - PowerPoint PPT Presentation

▶
ds595 cs525 urban network analysis urban mobility
SMART_READER_LITE
LIVE PREVIEW

DS595/CS525: Urban Network Analysis -- Urban Mobility Prof. Yanhua - - PowerPoint PPT Presentation

Sep 07, 2022 101 likes •433 views

Welcome to DS595/CS525: Urban Network Analysis -- Urban Mobility Prof. Yanhua Li Time: 6:00pm 8:50pm Wednesday Location: Fuller 320 Spring 2017 2 A few things Team assignment Finalized. (Great!) Guest Speaker 2/22 Affect a

slide-1
SLIDE 1

DS595/CS525: Urban Network Analysis

  • -Urban Mobility
  • Prof. Yanhua Li

Welcome to

Time: 6:00pm –8:50pm Wednesday Location: Fuller 320 Spring 2017

slide-2
SLIDE 2

2

A few things

  • Team assignment
  • Finalized. (Great!)
  • Guest Speaker 2/22
  • Affect a bit of presentation schedule
  • Merge two classes DS595 and CS525 in myWPI
  • Project 1
  • Proposal due today
  • A reminder of Project 1 follow-up.
  • Class website - project page for the timeline
slide-3
SLIDE 3

3

How to write good paper reviews/critiques

  • Summarize the work (80 points)
  • What problem?
  • Why the problem is important?
  • Which method is proposed?
  • How is the work evaluated?
  • Correctly summarize all these in your own words,

you get 80 points

  • Critiques/comments (20 points) Quality Matters
  • Some critiques, something in the paper is wrong?
  • Some future work to make the work more solid?
  • Some changes to the method to enable better

performances?

?

slide-4
SLIDE 4

4

Weka Online Website

  • 3 Volunteer Groups

Class 1,2 - Getting started with Weka, Evaluation Class 3 - Simple classifiers WEEK 4 Class 4 - More classifiers WEEK 5 Class 5 - Putting it all together WEEK 6 10-15 MINUTES EACH SESSION at the beginning of the class Briefly introduction of techniques you learned Post a question to the audience.

?

slide-5
SLIDE 5

Mining Spatio-Temporal Reachable Regions over Massive Trajectory Data

Guojun Wu#, Yichen Ding#, Yanhua Li#, Jie Bao†, Yu Zheng†, Jun Luo*

#Worcester Polytechnic Institute (WPI) †Microsoft Research,

*Shenzhen Institutes of Advanced Technologies

slide-6
SLIDE 6

Big Trajectory Data in Urban Networks

Taxi GPS Trajectory Mobile User Trajectory

  • Urban roving sensors deliver big trajectory data.
  • Reveal moving patterns and urban issues.

How to manage and utilize the big trajectory data to improve people’s life quality? Challenge

slide-7
SLIDE 7

7

Reachability Query

10 mile Free Space Road Network ST Reachability State of Art Our Proposed Query 10 mile 15 min

slide-8
SLIDE 8

8

Terminology

  • Trajectory:
  • a sequence of spatio-temporal points.
  • (traj_ID, latitude, longitude, timestamp, travel speed,

direction, occupancy).

  • Trajectory reachability:
  • Given S, T, L, r, tell if r are reachable from S in [T,T+L]
  • Reachable area:
  • Given S, T, L, find all {r} that are reachable in [T,T+L]
  • Prob-reachable area:
  • Given S, T, L, find all {r} that are reachable in [T,T+L] at least

prob% of days in the past

slide-9
SLIDE 9

9

Spatio-Temporal Reachability Query

  • Definition:
  • To find the reachable area in a spatial network from a

location in a given time period.

  • Example:
  • Start from my home at 8:00AM, where I can reach in 30

minutes with more than 80% confidence ( a ) ( b )

slide-10
SLIDE 10

10

Real-World Problems

6 pm 1 pm

( b ) Location-based Advertising

6 pm 1 pm

( a ) Location-based Recommendation

6 pm 1 pm 6 pm 1 pm

( c ) Business Coverage Analysis ( d ) Emergency Dispatching Analysis

?

slide-11
SLIDE 11

11

Exhaustive Search

  • Start from the querying location S and time T, to

search the neighboring road segments throughout the whole road network.

  • Long responding time for large trajectory dataset
  • In Nov 2014, Shenzhen, China;
  • Query: Reachable region from a user-specified

location and time to travel 1 hour, with a confidence of 80% or more. Taxi GPS 1.58 TB 22,083 taxis Bus GPS 1.34 TB 8,427 buses 10 minutes to get the query answers!

slide-12
SLIDE 12

12

Query Processing Framework

Lng Lat Time

S T-Index Con-Index

  • 1. Preprocessing
  • 2. Index Construction 3. Query Processing

S, T, 𝑀, 𝑄𝑠𝑝𝑐

Map- Matching

Trajectory Database

Road Segments

Mapped Trajectory

Input Clean

Road Networks

𝑄𝑠𝑝𝑐

r3 r2 rn r1 r3 r2 rn r1 r3 r2 rn r1

𝑀

Time

slide-13
SLIDE 13

13

Preprocessing

  • Map Matching
  • Map trajectory point to

real road network

  • Road Re-segmentation
  • Partition the original road

segments based on the given spatial granularity, e.g., <=500 meters

slide-14
SLIDE 14

14

Cities OS People The Environment Win Win Win Urban Computing

Tackle the Big challenges in Big cities using Big data!

Urban Sensing & Data Acquisition

Participatory Sensing, Crowd Sensing, Mobile Sensing Traffic Road Networks POIs Air Quality Human mobility Meteorolo gy Social Media Energy

Urban Data Management

Spatio-temporal index, streaming, trajectory, and graph data management,...

Urban Data Analytics

Data Mining, Machine Learning, Visualization

Service Providing

Improve urban planning, Ease Traffic Congestion, Save Energy, Reduce Air Pollution, ...

Urban Computing: concepts, methodologies, and applications. Zheng, Y., et al. ACM transactions on Intelligent Systems and Technology.

slide-15
SLIDE 15

15

Indexing Structure

  • ST-Index

B-Tree R-Tree

slide-16
SLIDE 16

16

Indexing Structure

  • Connection Index
slide-17
SLIDE 17

17

Query Processing

  • Single Location
  • Find maximum

bounding region

  • Trace back

search

r*

slide-18
SLIDE 18

18

Query Processing

  • Multiple Locations
  • Find unified maximum bounding region
  • Trace back search

r1 r2 r3 r4 r1 r2

slide-19
SLIDE 19

Evaluation

  • Dataset: 60GB real taxi mobility data in Shenzhen
  • Baseline Algorithm

– Exhaustive search

  • Evaluation metric

– Running time (s) – Total Length of Road Segments (km)

Statistics Value City Size

400 square miles

City Taxi Population

21,358

Duration

November 2014

# of spatio-temporal points

400 million (407, 040, 083)

slide-20
SLIDE 20

20

Evaluation

90% 50% Running time: 50-90% reduction over ES Road Segment Length: Increases as L increases

slide-21
SLIDE 21

21

Evaluation (T)

slide-22
SLIDE 22

22

Evaluation (Prob)

slide-23
SLIDE 23

23

s-query vs m-query

3 times Running time: 3 times reduction over s-query Running time: Constant vs linear

slide-24
SLIDE 24

24

Evaluation (m-query)

slide-25
SLIDE 25

Summary

  • Approximate query processing

– Single trajectory aggregate query

  • via Random Index Sampling (RIS)

– Concurrent trajectory aggregate queries

  • via Concurrent Random Index Sampling (CRIS)
slide-26
SLIDE 26

26

Dimensions of Query

Data Static Streaming Mode Local Distributed Spatial Single location Multiple location

  • Spatio-Temporal Reachability Queries have different

types regarding different data inputs

Union Join Sequential

slide-27
SLIDE 27

27

Dimensions of Query

Union Join Sequential A B C

slide-28
SLIDE 28

28

Dimensions of Query

  • Streaming Data
  • Real-Time Problem
  • Distributed Mode
  • Large-Scale Database
  • Concurrent Queries

6 pm 1 pm

Emergency Dispatching Analysis

slide-29
SLIDE 29

Thank you!

slide-30
SLIDE 30

30

Predictive Query

  • Transitive Reachability
  • AàB, BàC AàC
  • Bayesian Network
  • The probability that an object travel from segment

A to segment B based on speed information

slide-31
SLIDE 31

Imagine This…

  • You get an offer from company X and you need to find

where to live

  • House A, 8 miles to company, 15 min.
  • House B,10 miles to company, 10 min.
  • Which one?