# Visual Analytics Course

## Visual Analytics Course

Instructors: Guy Melançon (email: Guy dot Melancon at labri dot fr), Bruno Pinaud (email: Bruno dot Pinaud at labri dot fr)

### Bordeaux roadmap 2012-2013 / 1st Homework

Remarks on homework 1. Many (many) of you made the same mistakes, going forward on personal comments, collecting impressions rather than offering a well structured and a full account of all positive aspects and flaws of the visualization you studied. Criticizing the work of others must be done on objective criterias, it's not about giving impressions and throwing I like/I don't like judgements.

So what do you need as base material before you can judge and criticize a piece of work? Criterions. Your first course did lay down some criterions that can be used to evaluate the efficiency of a map. It's all about distances. And because the core intent of a map is to carry information (even some knowledge), we may judge it as being a media conveying a message. So our judgement should be about how well the map does convey that message.

So when criticizing a map, you should

1. Identify the message the map is trying to convey
2. And then explain whether the map does a good job at doing / not doing it

The methodology we suggested is based on a distance taxonomy and is borrowed from a paper by paper by Nanard & Nanard (2002). Intuitively, the efficiency of a message is measured in terms of distance between the issuer of the message and the receiver. The smallest the distance, the better the media was at conveying the original message. This idea, we owe to communication theory, actually has its place in a broad range of discipline. When it comes to measuring this distance in graphical user interfaces – which is how we look at a map, the distance is broken down into three distinct components.

• Operational distance: this refers to graphical items affecting the immediate perception of the map (the right term is 'pre-attentive', things that require no cognitive efforts, nothing you consciously do). Color choices, color contrast, physical distribution of graphical elements on the screen, etc. The chapter on perception investigates items that are pre-attentively processed.
• Articulatory distance: now, this has to do with the ease (or difficulty) a user may have to get the right information. There might be a legend on the map, but you may need to move your mouse over to unroll it – a better option might be to have it easily accessible at all times. The map designer may have used pop-ups to bring information upfront, but then this information may make the map partly invisible which may impair readability. Observe how this does not rely on a pre-attentive process. These are graphical (or GUI) elements that are offered to the user to help the map convey its message.
• Semantic distance: this has to do with meaning. Is the colormap adequatly reflecting the underlying information it carries. Typically, you would use a red/green pattern to indicate a good/bad opposition (note how this may depend on cultural conventions or habits). using a rainbow colormap to display quantitative information is bad, using a light to dark scale is much better. Note again that what we are judging here is not whether the colormap has been properly embedded in a legend or not, we are judging its efficiency at carrying the semantics of the map. Similarly, we are not judging whether the color contrats of the colormap is right or wrong – this has to do with pre-attention.

All in all, these criteria should help you fill up an evaluation grid like:

So, to finish up this (rather long) comment. Most of you did not use such a rigourous and objective grid to evaluate your map, as was expected.