Talk:Lesson05

10 Lessons in Treemap Design

This article really strikes me. I have to disagree with some of the treemap designs that the author has stated in the treemap. For the 2nd point where he stated that space matters, the author stated that having category would distort labelling. I disagree with the author, and further think that It would in fact aid the user. The picture illustration that he gave shows a white borders to group a category together. The picture actually gave me a clearer indication of the various groups and do not distort the sizes. This was due to the (different coloured) borders showing me a clear indication of the different groups. Another example of a good use of spacing between the various categories is a graphic made by the New York Times. http://www.nytimes.com/imagepages/2007/02/25/business/20070225_CHRYSLER_GRAPHIC.html

For the 4th point, the author stated that the labels must stand-out against treemap colours. This statement is dependent on the colours and how the labels are used. Even if you use colours that would contrast with the treemap colours and the labels. If everything is cluttered as shown in the example, it would still make the diagram look messy and undesirable. Lastly, creating a glow around the label as stated by the author may not always work.

-Ong Ming Hao

Treemap in D3

The following blog shows how treemap is done by D3. The cool thing about this web is that the treemap on the web is as interactive as the tree map in Tableau. In addition, the transition between selections is amazing, showing the resizing of the boxes. http://bl.ocks.org/tgk/6044254 Thinking that this might be useful for some of our projects.

Treemap is definitely a powerful analytical technique. However, initially it striked me as too complicated. Only after listening to prof., I started to realize the power of it. Nevertheless, without much previous experience and interactions, users might just be confused. Hence, when we intend to use such a powerful tool, keep in mind to add legend explaining what represents which. For instance, the bigger the size, the larger the sales. And the darker the color, the higher the price. Tableau seems to miss out the size legend. Here is an example of how NYTimes defines the meaning of the box size as well as the color of its treemap. http://www.nytimes.com/interactive/2008/09/15/business/20080916-treemap-graphic.html

-Nguyen Duy Loc

US Economic Census Treemap

The following article demonstrated the use of Treemap to show insights from the data on the US Economy and briefly mentioned applications which can draw Treemap. The article also proposed useful various hierarchical structures for different contexts. For example, when visualizing student performance data, the following hierarchical structure is proposed

• Subjects -> Units -> Lessons

In this particular context, the classroom time can be illustrated with the size measure while the grade can be illustrated with the color measure. Thus, it provided a good starting point in terms of initiating the analytical process for beginners like us.
The author then used data from 1997 to 2002 US Economic Census and illustrated it in a Treemap. The use of the Treemap in this context was an intelligent one. At one glance, the reader is able to draw 2 quick facts with regards to the US economy from 1997 to 2002. Firstly, the reader can know that Health Care and Social Assistance, Retail Trade and Manufacturing sectors are the top 3 contributors to the US economy. Secondly, the reader is able to understand that the Manufacturing sector has been facing a decline over the years.
However, there is one feedback which I have on this particular Treemap. Some of the labels are not fully printed, instead, some of them were truncated and replaced with “…”. This will cause poor readability of the chart. As Prof had mentioned in the Webex, we should avoid truncation of the labels in data visualization. There is no meaning if the reader is unable to understand what the label is trying to say. This will only confuse the reader. Thus, in light of such a situation, we should avoid printing the label. Alternatively, we can resize the entire chart so that the major/all labels can be printed on the chart (this may not be applicable in some cases).
-Tan Kee Hock

FundExplorer

FundExplorer is a project which aims to help investors make informed decisions on the diversification of their mutual fund portfolio. This project utilises the concept of a Context Treemap, which is another kind of treemap. In Prof’s remote lesson, we learnt what a treemap was and how it was used to visualise multivariate data hierarchically.

This concept of a Context Treemap was particularly interesting to me as, instead of neatly placed square areas that represent values in a hierarchical data set, the Context Treemap shows “a distorted treemap view in which 0-valued items are given some proportion of the overall display area”. Simply put, variables with values that are 0 will still be shown on the treemap, thus making the treemap look a little pock-marked. In terms of stock portfolio, one benefit the Context Treemap claimed to have over the conventional treemap was that it preserves the proportions that were held for stocks. Furthermore, the visualization of data within its context enables investors to visually determine the diversification of their portfolio within the stock market.

I do agree that this project gives another insight into how proper visualisations can add another dimension into an industry which typically does not use visual analytics in their decision-making process. I feel that this could even make decision-making for mutual funds (and investments in general) more accessible for those who are intimidated by the many ratios and calculations one may need. However, as what the paper says, I do have my doubts on the efficacy of this “new” type of visualisation. The results of the visualisation do not improve the visual clutter that other visualisations also produce. Furthermore, this is also not a one-stop solution for making decisions on how to diversify a portfolio, as the task of diversifying a fund portfolio is usually part of a larger task that aims at balancing the risk-profit-ratio.

What do the rest think about this project and in what other industries have you seen visual analytics being used (that you may not have expected)?

- Cornelia Tisandinia Larasati

Sunburst Diagram Techniques

This article suggested several ways to overcome observers feedback that the small “slice” size of files and directories that were near the periphery of the visualization and said that it was difficult to distinguish the different attributes.

Angular Detail Method The selected item appears to extend out of the overview, and then it and its children expand radially outward to occupy a larger display area.

Details Outside Method When a viewer selects an item, the entire hierarchy shrinks in the center of the display and stays there. The selected item emerges fromthe overview on the edge of the overview closest to its position, and the item expands to be a new complete circular ring-shaped region around the overview.

Details Inside Method Similar to the Details Outside Method, the selected item then extends inward to the center of the window, is drawn as a circle, and expands radially to occupy the center of the image.

- Alson Tan

Choosing between Treemap and Sunburst

Both are great visualization techniques used to display hierarchical data. There are modern techniques that complements or strengthens existing charts like bar, pies and line graphs.

But when do we use one over the other? The answer lies in the information you want to convey in your charts.

Treemaps by with their rectangular shape are better suited for comparison across the hierarchical level. Human minds differentiate with ease visual items such as size, shape and color. This is what Treemap does really well. And by fully utilizing the space given, Treemap manages to display much more data across a hierarchy than a Sunburst ever could.

In contrast, Sunburst charts are circular in shape and harder to differentiate in size as compared to rectangles and straight lines. The circular shape also mean much of the space to tell your story is lost due to poor utilization of areas. But what Sunburst excels in, is painting a much fuller picture than Treemap by illustrating the different depths of hierarchical groups. Exploration of a certain slice of the Sunburst chart reveals deeper insights and draws attention to the main message you want to get through to the reader.

I hope this gives you a quick insight when choosing between these 2 really beautiful visualizations. But most importantly, try both visualization charts out and see, if the story you want to tell, is delivered best through which design.

- Tee Yu Xuan