Applications of Spatial Data Mining & Visualization - Case Studies.

Slides:



Advertisements
Similar presentations
Decision Support and Artificial Intelligence Jack G. Zheng May 21 st 2008 MIS Chapter 4.
Advertisements

Pseudo-Relevance Feedback For Multimedia Retrieval By Rong Yan, Alexander G. and Rong Jin Mwangi S. Kariuki
V-1 Part V: Collaborative Signal Processing Akbar Sayeed.
Kien A. Hua Division of Computer Science University of Central Florida.
Content-Based Image Retrieval
Mr. Burton 1.2 notes Please Grab: 1. Your folder. 2. Writing Utensil. 3. Answer the following question: Please write down what you feel are the FIVE themes.
Raster Based GIS Analysis
Chapter 1 – Uncovering the Past
Border around project area Everything else is hardly noticeable… but it’s there Big circles… and semi- transparent Color distinction is clear.
ADVISE: Advanced Digital Video Information Segmentation Engine
Supervised by Prof. LYU, Rung Tsong Michael Department of Computer Science & Engineering The Chinese University of Hong Kong Prepared by: Chan Pik Wah,
Self-Supervised Segmentation of River Scenes Supreeth Achar *, Bharath Sankaran ‡, Stephen Nuske *, Sebastian Scherer *, Sanjiv Singh * * ‡
Vision Computing An Introduction. Visual Perception Sight is our most impressive sense. It gives us, without conscious effort, detailed information about.
Dieter Pfoser, LBS Workshop1 Issues in the Management of Moving Point Objects Dieter Pfoser Nykredit Center for Database Research Aalborg University, Denmark.
©2005 Austin Troy. All rights reserved Lecture 3: Introduction to GIS Part 1. Understanding Spatial Data Structures by Austin Troy, University of Vermont.
CHAPTER 6 Statistical Analysis of Experimental Data
What is our weather like? We are learning why there are differences in the weather in Britain.
Dr. David Liu Objectives  Understand what a GIS is  Understand how a GIS functions  Spatial data representation  GIS application.
Emotional Intelligence and Agents – Survey and Possible Applications Mirjana Ivanovic, Milos Radovanovic, Zoran Budimac, Dejan Mitrovic, Vladimir Kurbalija,
Introduction to machine learning
Studying Geography The Big Idea
MDSS Challenges, Research, and Managing User Expectations - Weather Issues - Bill Mahoney & Kevin Petty National Center for Atmospheric Research (NCAR)
Recent polls have shown that 1/5 of Americans can’t locate the U. S
Image Classification and its Applications
Meteorological and Hydrological Service, Grič 3, HR Zagreb, Croatia FORECASTING BORA WIND AFTER THE COLD FRONT PASSAGE AT.
Computer vision.
Overview Dennis L. Johnson What is GIS? Geographic Information System Geographic implies of or pertaining to the surface of the earth Information implies.
How Geographers See the World
Chapter 2 Section 3 Winds.
Data Management Turban, Aronson, and Liang Decision Support Systems and Intelligent Systems, Seventh Edition.
Ihr Logo Chapter 5 Business Intelligence: Data Warehousing, Data Acquisition, Data Mining, Business Analytics, and Visualization Turban, Aronson, and Liang.
Point to Ponder “I think there is a world market for maybe five computers.” »Thomas Watson, chairman of IBM, 1943.
CS654: Digital Image Analysis Lecture 3: Data Structure for Image Analysis.
Please turn your syllabus in to the bin.
Spatial Data Analysis Yaji Sripada. Dept. of Computing Science, University of Aberdeen2 In this lecture you learn What is spatial data and their special.
Time Series Data Analysis - I Yaji Sripada. Dept. of Computing Science, University of Aberdeen2 In this lecture you learn What are Time Series? How to.
Themes & Essential Elements. Human Geography Studies distribution and characteristics of the world’s people (where people live and what they do) Examines.
Data Warehouse. Design DataWarehouse Key Design Considerations it is important to consider the intended purpose of the data warehouse or business intelligence.
Modeling Storing and Mining Moving Object Databases Proceedings of the International Database Engineering and Applications Symposium (IDEAS’04) Sotiris.
Chapter 1 – A Geographer’s World
An Internet of Things: People, Processes, and Products in the Spotfire Cloud Library Dr. Brand Niemann Director and Senior Data Scientist/Data Journalist.
1 Introduction to Software Engineering Lecture 1.
Ihr Logo Chapter 5 Business Intelligence: Data Warehousing, Data Acquisition, Data Mining, Business Analytics, and Visualization Turban, Aronson, and Liang.
The ISTE National Educational Technology Standards (NETS  S) and Performance Indicators for Students.
Chapter 3. Information Input and Processing Part – II* Prepared by: Ahmed M. El-Sherbeeny, PhD *(Adapted from Slides by: Dr. Khaled Al-Saleh) 1.
King Saud University College of Engineering IE – 341: “Human Factors” Fall – 2015 (1 st Sem H) Chapter 3. Information Input and Processing Part.
1 Pattern Recognition Pattern recognition is: 1. A research area in which patterns in data are found, recognized, discovered, …whatever. 2. A catchall.
Chapter 5: Business Intelligence: Data Warehousing, Data Acquisition, Data Mining, Business Analytics, and Visualization DECISION SUPPORT SYSTEMS AND BUSINESS.
Unit 2: Geographical Skills
The Six Essential Elements of Geography. What is Geography?  The study of the physical, biological & cultural features of the Earth’s surface.
Geographical Data and Measurement Geography, Data and Statistics.
2004 謝俊瑋 NTU, CSIE, CMLab 1 A Rule-Based Video Annotation System Andres Dorado, Janko Calic, and Ebroul Izquierdo, Senior Member, IEEE.
Distributed Data Analysis & Dissemination System (D-DADS ) Special Interest Group on Data Integration June 2000.
Using Bayesian Networks to Predict Plankton Production from Satellite Data By: Rob Curtis, Richard Fenn, Damon Oberholster Supervisors: Anet Potgieter,
Fast SLAM Simultaneous Localization And Mapping using Particle Filter A geometric approach (as opposed to discretization approach)‏ Subhrajit Bhattacharya.
CLUSTERING GRID-BASED METHODS Elsayed Hemayed Data Mining Course.
MODULE 4 1Module 4: Effects of Climate Change What are the risks of a changing climate?
A person who studies people, events and ideas of the past. Who is a historian?
World Geography Chapter 1. The Study of Geography Section 1.
1 INTRODUCTION TO COMPUTER GRAPHICS. Computer Graphics The computer is an information processing machine. It is a tool for storing, manipulating and correlating.
The PLA Model: On the Combination of Product-Line Analyses 강태준.
SmartMet Lea Saukkonen FMI What is SmartMet? A software tool for visualizing and editing meteorological data.
Chapter 1 – A Geographer’s World
Ch 1 A Geographer’s World
Your Name METR 104 Today’s date
Your Name METR 104 Today’s date
Maps.
Point Cloud Processing
Presentation transcript:

Applications of Spatial Data Mining & Visualization - Case Studies

2 Introduction Meteorological Data and Demographics Data hold important information that can help in several application contexts Several data mining applications possible on these data sets In the department we have research projects working on these data –RoadSafe – Summarizing large spatio-temporal weather prediction data –Atlas.txt – Summarizing UK 2001 Census data Both these projects present summaries to users in natural language, English and other modes Real World applications contain data mining as one of the modules or tasks in the project –Not as the end product in itself

3 Road Ice Forecasts -RoadSafe Road Ice Forecasts: –Are required by local councils for winter road maintenance operations –Are driven by computer simulation models that predict weather conditions for 1000’s of points on a road network –Output of model is a huge spatio-temporal data set (up to 33mb for some councils) –Form part of a road forecasting service delivered to Road Engineers via an online Road Weather Information System (RWIS) RWIS allows model data to be communicated in various modalities, e.g. text, tables, graphs and maps

4 Model output is a large spatio-temporal data set (in order of Megabytes) Road network split into routes, 9 meteorological parameters (e.g. Road Surface Temperature) measured at each point on a route Sampled at 20 minute intervals over a 24hr period

5

6 24 Hour Forecast for Kirklees All RoutesMin RST Time <= 0c Ice Hoar Frost SnowFogMaxGustsRainTS Worst/Best-1.1 /1.421:00 /NAYes /NoNo/No Yes/Yes15/13No /NoNo Wind (mph) Light south to south-easterlies for the duration of the forecast period. Winds may become more moderate late morning on higher ground, but remaining southerly. Weather A mainly cloudy night, with foggy patches across much of the forecast area. Higher ground above the low cloud level could see temperatures drop below freezing during the late evening, with most western parts of the forecast area dropping below freezing by the morning. Urban areas are expected to remain marginal throughout the night. RouteAll routes summary worst/best 10.4/1.8NA/NANo/No Yes/Yes13/11No /NoNo 20.7/2.0NA/NANo/No Yes/Yes13/10No /NoNo 30.5/1.8NA/NANo/No Yes/Yes13/9No /NoNo 40.4/1.8NA/NANo/No Yes/Yes13/12No /NoNo 50.7/1.9NA/NANo/No Yes/Yes13/9No /NoNo 60.7/2.1NA/NANo/No Yes/Yes13/11No /NoNo 70.9/1.8NA/NANo/No Yes/Yes13/9No /NoNo 80.8/2.1NA/NANo/No Yes/Yes13/9No /NoNo 91.4/2.1NA/NANo/No Yes/Yes13/9No /NoNo 100.8/1.9NA/NANo/No Yes/Yes13/9No /NoNo 110.3/1.8NA/NANo/No Yes/Yes13/11No /NoNo /1.522:40 /NAYes /NoNo/No Yes/Yes15/11No /NoNo

7 Problem Input: Spatio-temporal weather prediction data (shown on slide 4) Output: Summary of input data (shown on slide 6) Task:? –There is no well defined data mining task (classification or clustering or a new task) –Clusters of similar weather spatially and temporally can be one kind of summary –Classification of routes can be another kind of summary –Both used in the final system Challenges –Complex spatio-temporal data set –Spatio-temporal analysis methods are still maturing –Even visualization of the entire data is hard

8

9 Overview of Data Analysis Two main challenges: –Analysing the input data along the temporal dimension –Analysing the input data along the spatial dimension Ideally analysis should be performed on both dimensions simultaneously Solution inspired by Video Processing –The input data set is seen as a video containing 3*24*9=648 frames (maps) 3 key elements: 0. Pre-processing – geo-characterization – merging required data with other relevant themes 1.Low level processing -Global Trends – Temporal segmentation -Local Events – Spatial Segmentation (Classification and Clustering) 2.Event detection and indexing 3.Keyframe extraction. Extracted keyframes form the summary

10 Preprocessing Frames of reference used for spatial clustering Geographic Characterisation assigns properties to each data point based on frames of reference for the region

11 Spatial Reference Frames Spatial descriptions should be meteorologically correct (not necessarily most geographically accurate) Forecasters consider how geography influences weather conditions in their descriptions (meteorological inferences) "exposed locations may have gales at times” Dominant geographical features within regions also affect the reference strategy Kirklees (land locked) Hampshire 1.Altitude 1. Coastal Proximity 2.Direction 2. Altitude 3.Population 3. Direction 4. Population

12 Spatial Segmentation Each of the 648 frames (maps) are analysed to compute spatial segmentations (clusters) Because weather parameters are continuous, they are first discretized E.g for road surface temperature (map shown on the next slide) –OK => {>4} –Marginal => { 1} –Critical => { 0} –Subzero => {<=0} Density based clustering used for performing spatial segmentation

13 Discretization of weather parameters

14 Cluster Densities Frame of Reference Proportion of subzero points 07: :00 08:20 08:40 Altitude 0m: m: m: m: m: m: Direction Central: Northeast: Northwest: Southeast: Southwest: Urban/Rural Rural: Urban:

15

16 Atlas.txt Is an ongoing research project –Produces textual summaries of geo-referenced statistics –for visually impaired users The focus of the project is more on visualization of spatial data by visually impaired (VI) users –Spatial data is essentially geometric and it is not clear how visually impaired users model geometric information –In the absence of vision, is it possible to model geometric information based on tactile and audio inputs? If possible, what is the nature of these mental models of geometries

17 Input %Unemployment in Aberdeen <2.2 <3.5 <4.8 <6.1

18 Output No gold standard models of spatial information suitable to VI users available So several alternative summaries of spatial information that need to be tested on real users One possible example textual summary: “Some wards in the east and central parts (3,5,6,9) of the city have high percentage of unemployed people aged above 03.51%” Are the textual summaries adequate on their own? Do they need to be supplemented by tactile or sonic maps? –Tactile maps –Sonic Maps

19 Problem Input: 2001 UK census data Output: Summary of input data Task: Spatial segmentation + Spatial visualization for VI users –Unlike RoadSafe the data mining task is well defined –What is less defined though is the task of visualization of summary by VI users –Shape (geometry) and topology of segments need to be accessible to visually impaired users

20 Space and Visual Impairment Atlas.txt is an ongoing research project –more open questions than useful answers VI users need to perform two tasks for modeling spatial data –Scanning space for information Several scanning strategies possible E.g. Left-right VS top down –Coding spatial information using a suitable reference frame Once again several coding strategies available E.g. body (ego) centric VS external VI users are trapped in a vicious circle while finding efficient scanning and coding strategies

21 Strategic Disadvantage for VI users Scanning strategy determines the quality of spatial information acquisition –But better scanning strategy possible only with knowledge of spatial information Sighted users take a quick look at an image which helps them to scan the image lot more efficiently VI users do not have the luxury of a quick glance! Coding strategy determines the quality of mental representation –Mental models coded on body centric reference frame less useful for complicated spatial analysis –External reference frames help to code better quality mental models –VI users need improved scanning strategies for acquiring suitable external reference frames –Because VI users are disadvantaged to find a quality scanning strategy, they are also disadvantaged to find a quality coding strategy

22 Solution Options VI users clearly need external help in finding suitable external reference frames Atlas.txt solution –Identify several reference frames and present summary coded in each of these –VI users may be familiar with some spatial layouts E.g. telephone key pad and clock face –Use several of these to code summary information “Some wards in the east and central parts (3,5,6,9) of the city have high percentage of unemployed people aged above 03.51%” –E.G. ‘east and central parts’ can also be expressed by (3,5,6,9) each number referring to a location on the telephone keypad layout