Visual Analytics and Data Mining Visual Analytics and Data Mining in S- in S -T T- -applications applications Gennady Andrienko & Natalia Andrienko Fraunhofer Institute AIS Sankt Augustin Germany http://www.ais.fraunhofer.de/and Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 1 A View on S- -T Data Mining T Data Mining A View on S Complex and multidimensional; Data May contain errors Complexities: Input data 1) Space, May have many 2) Time, parameters; 3) Multiple May be computationally attributes & intensive Method(s) dimensions 4) Outliers, discontinuities Need to be interpreted; To be used for directing further analysis Output data Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 2 1
Complexities Complexities • Number of attributes • Length of time series • Number of spatial objects • High dimensionality • Abrupt temporal changes • Great variability of values Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 3 Complexities: example 1 Complexities: example 1 Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 4 2
Complexities: example 2 Complexities: example 2 Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 5 Aggregation method 1: by intervals Aggregation method 1: by intervals Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 6 3
Aggregation method 2: by quantiles quantiles Aggregation method 2: by Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 7 Attend to particulars: high variation Attend to particulars: high variation 1. Aggregate time graph by quantiles 2. Save counts 3. Visualise e.g. on a scatter plot 4. Select items with high variation Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 8 4
Attend to particulars: high fluctuation Attend to particulars: high fluctuation • Select items with maximal number of jumps between quantiles Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 9 Attend to particulars: stable extremes Attend to particulars: stable extremes • Select items being always in the topmost 10% Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 10 5
Attend to particulars: extreme changes Attend to particulars: extreme changes 1. Transform the time graph to show changes 2. Select extreme changes in a specific year (here 2003) Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 11 Visual Analytics in Data Mining Visual Analytics in Data Mining Data Data Result preview preparation exploration Method Method and selection application visualisation, data interpretation display manipulation manipulation, visualisation, etc. etc. Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 12 6
Requirements to Visual Analytics Requirements to Visual Analytics • Space- and Time-awareness • Work with complex multidimensional data • Support for uncertain and missing data • Scalability • Support and encouraging of several complementary views on the same data • Dynamic linking and coordination of several data displays • From the overall view to particulars of interest • From idea generation to hypothesis testing using statistical methods, followed by reporting Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 13 Potentially useful tools for MSTD Potentially useful tools for MSTD � Information visualisation tools, for example, HCE & TimeSearcher from HCIL, Univ. Maryland � Geovisualisation tools, for example GeoVistaStudio (Penn State Univ.) and Descartes/CommonGIS (Fraunhofer Institute AIS) � Graphical statistics tools, for example, Manet & Mondrian (Augsburg Univ.) � Usually such systems are research prototypes that implement innovative ideas, but provide restricted functionality and limited user support Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 14 7
Still open issues (for all tools!) Still open issues (for all tools!) � Work with qualitative (non-numeric) data � Work with fuzzy, uncertain, and missing data � Continue scalability efforts � Support in processing and management of findings: recording, structuring, browsing, searching, checking, combining, interpreting… � Help in visual communication of derived data, constructed knowledge, and recommended decisions � Adaptability to user, data, tasks, and hardware � Embedding intelligence into software for helping users and avoiding cognitive overload Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 15 EDA: from Practice to Practical Theory EDA: from Practice to Practical Theory � Data � Tasks � Tools � Principles to appear ≈ end 2005 Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 16 8
Visual Analytics & Data Mining Visual Analytics & Data Mining 1. Do they need each other? 2. How to benefit from combining two scientific disciplines and related technologies? 3. How to develop each of two scientific disciplines for achieving a synergy? Mining Spatio-Temporal Data @ PKDD, Porto, 2.10.2005 17 9
Recommend
More recommend
