1 / 37

Data Visualisation

HCI 0283 Lecture 5 Dynamic Exploration. Data Visualisation. Real Problems. Real world problems are very seldom precisely specified

ross
Télécharger la présentation

Data Visualisation

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. HCI 0283 Lecture 5 Dynamic Exploration Data Visualisation

  2. Real Problems • Real world problems are very seldom precisely specified • We want a house for under about £100,000 with at least three bedrooms (although four would be nice), but it’s got to be close to a good school and it would be quite nice if it was near Granny so she could baby-sit for us • Many other problems are specified just as imprecisely • The reason for this is a lack of knowledge • The house-hunter has no idea of • What houses are currently on the market • Where the local schools are • The general price trends • Local variations

  3. Real Problems • The first task is therefore to help the house-hunter understand the relevant data and any hidden relationship within it • The next task is to help the house-hunter solve his problem (finding a house) • During this process the house-hunter may change his requirements • It looks like we might be able to afford a 5-bedroom house if we rent out a couple of rooms • Formulating the problem is as important as finding a solution

  4. Command Line Queries • Conventional databases require queries to be entered in very specific manners • SELECT house-address FROM mydatabase WHERE price < 100,000 AND bedrooms >=3 • The result of this might well be • 0 records returned • 1433 records returned • Neither of these is particularly useful

  5. Command Line Queries • There are many problems with conventional database query languages • Learning a new language • Errors are not tolerated • Too few or too many records returned • No suggestions how to reform a query • Slow! • No context data • Difficult to build a cognitive map of the data

  6. Command Line Queries • The assumption behind command line queries is that the user knows precisely what question he wishes to ask • In fact this is very rarely the case • Dynamic exploration allows the user to continually reformulate his question in light of the information presented to him.

  7. Dynamic Queries • We could replace the command line query with a dynamic display Price Bedrooms Travel time

  8. Dynamic Queries • This makes the task a ‘what if…?’ activity • He is able to explore the data more easily by restating the question in more general terms • Given a collection of objects, each described by the values associated with a set of attributes, find the most acceptable such object or, perhaps, a small number of candidate objects worthy of more detailed consideration • It no longer matter what you are looking for, the important thing is that you can explore the data through a graphical interface

  9. Dynamic Queries • A similar type of task is the selection of a film to watch on DVD • In FilmFinder each film is represented by a coloured square, the colour representing the genre (SF, horror, action, comedy etc) • Alphasliders are used to select subsets of films based on dates, actors, directors or other attributes

  10. Alphasliders • Alphasliders allow users to scan rapidly through and select items from list of alphabetical data Text output Slider thumb Slider area Hunt for Red October A B C D E F G H I J L M N O R S T W Z Index

  11. Attribute Explorer • A drawback to the dynamic queries approach is that data is only disposed when it satisfies the query • There are situations where the display of all data is useful to provide contextural information • This extra information may provide useful hints for exploration • In lecture 3 we showed that data in a histogram can be viewed interactively using sliders • This approach is extended using IBM’s Attribute Explorer to present data in a highly dynamic manner • Suppose we have car data for city MPG, highway MPG, engine size and horsepower

  12. Attribute Explorer

  13. Attribute Explorer

  14. Attribute Explorer

  15. Summary Information • It is not easy to display summary information for a histogram • One way is to show the mean value with a circle • The circles on each histogram move in response to each other

  16. Boolean Operations • Attribute Explorer allows the user to identify those cars which satisfy limits on city MPG AND highway MPG AND horsepower AND engine size • This is clearly a Boolean operation • The InfoCrystal allows a number of different Boolean operations to be performed on a dataset

  17. InfoCrystal

  18. Link Crystal • This approach is used as the Link Crystal within Attribute Explorer • E.g. an link crystal showing house price, number of bedrooms and garden size Price Bedrooms Garden size

  19. Fuzzy Queries • Mathematical and computing techniques exist to handle fuzzy logic • These may be useful, but are generally designed to solve specific optimisation problems • Our selection problem is not very specific! • The best fuzzy computer is still the human brain

  20. Coffee Time!

  21. Very Large Databases • The only limit to the number of attributes that can be handled by Attribute Explorer is the size of the display screen • Hypervariate data is always a problem • One technique is used in the VisDB tool • This uses vertical scales to select the attribute limits and allows the importance of the attributes to be weighted

  22. Very Large Databases • The objects are now ordered depending upon how well then satisfy the attribute values selected • The data is then arranged in a spiral pattern

  23. Neighbourhood Explorer • Neither Attribute Explorer nor VisDB have any mechanism for displaying pictures of objects • In some cases images are very helpful • If we are looking for a house then we will be comparing a relatively small number of houses, each of which is a close neighbour (in terms of attributes) of the others • Instead of arranging the attributes in a linear manner, they are arranged radially with the currently examined house at the centre • Houses appear on all attribute axes • A house is selected by dragging it to the centre, at which point all of the axes are rearranged

  24. Rooms Area Price Travel time Time

  25. Musical Visualisation • Many fields use symbolic notations that are unfamiliar to non-experts • Music • Choreography • Mathematics… • Musical notation makes it relatively simple to read a piece of music but more difficult to compose music • Often music is composed on an instrument then transcribed into musical notation • It is now possible to play an instrument and have a computer automatically convert it into a paper-based notation that other people can read • Even so, there are a number of different musical notations in use today…

  26. MusicNotations AmazingGracein ‘standard’ notation Thesamephraseinnumericalnotation

  27. Tablature for stringedinstruments LachrymaebyJohnDowlandinlutetab StairwaytoHeavenbyLed Zeppelin inguitartab

  28. MusicVisualisations • Readinganymusicalnotation is a skillthathastobelearned – otherwiseyoucanseeifthemusicgoesupordown, butthat’s aboutit • Anotheralternative is toshowasananimation not onlythe ‘upanddown’ of thenotesbutalsothe ‘length’ • Eg, http://www.youtube.com/watch?v=ipzR9bhei_o

  29. Musical Visualisation • To make composition easier for lay people instruments have been designed that give visual feedback as well as aural feedback • These encourage direct manipulation of objects in order to control the sound that is produced • http://www.grotrian.de/spiel/e/spiel_win.html

  30. Choreography Waltzinsquare, man’s steps LaCachucha, byFriedrich Albert Zorn. Beauchamp-Feuillernotation for a Frenchcourante

  31. Choreography

  32. Choreography MarianelaNuñez and Yohei Sasaki in a position from the 'Swan Lake' Act I pas de trois and the Benesh notation for the same position

  33. Summary • Real-world problems are seldom specified exactly and involve examining a local neighbourhood of a domain • This means that command line languages like SQL are often not very useful • Users need to be able to use dynamic queries and get rapid visual feedback • In order to be visualised and explored the data has to be represented in some symbolic method; some fields have very specialised symbolic notations

  34. Coming Soon… • Next lecture: Internal Models • Homework: Read chapter 5 of Information Visualisation (Spence)

More Related