1. I saw this Viz on World War I in the Tableau forum and without a doubt, I nominated it for Viz of the Day by sending a message to public@tableau.com on 7/22. In the message, I didn't write anything but the link to the viz. On 7/25, it was selected as the Viz of the Day. Note that between 22 and 25, there was a weekend.

    So, that's how good the viz is. I had nothing to add. The only concern I had was, the reference line will fall off the chart after 2018. The author Ryan Rowland replied that he will come up with a solution between now and 2018. I bet he will.

    A few days ago, I felt that for an information viz, it would be really helpful to the viewers if we add some web extensions to it for people to further explore. A good viz which can only host some brief introduction after all, must serve as a springboard to more discovery on the topic. I would add a link from each battle to Google or Wikipedia, like what I did in this viz.

    To my surprise, after click on the data marks, I found the links are there already in the tooltips. Ryan has thought about it already! Alas, viewers have to click or select the data marks to see the link.

    Before Select
    After Select
    So here comes an obscure feature in Tableau regarding tooltips: Show Tooltips on Hover.
    By turning it on, we can see the URL right away just by hovering, without click or select. Note that it is not the default setting. We have to turn it on manually. (Maybe I should send a request to make it the default. Or the text should be "Show URL/Link on hover"?)

    Try this tweaked version to see the effect. This concludes the #TweakThursday.
    0

    Add a comment

  2. Here I will present an alternative solution to Tableau's Story Points. The main difference is that it can accommodate global filters. This feature has been asked by many users in the Tableau forum. Besides, the formatting option is much limited in Story Points. More have been asked for. For example, people want to put a logo on top of the board. Here is Johan's collection of feature requests on Story Points.

    Here is an example in which we have 7 slides or story points. Click image to view the interactive version.
    The main design steps are:
    1. Create internal index fields
    Each index is a calculated field such as Point 1-7. The formula is of value 0. Can't be simpler.
    2.Create the tab sheet
    All the 7 tabs will be in one sheet. Right click the above index fields and drag them to the Columns shelf. Select Min() as the aggregation.
    The 7 Min() fields will create 7 columns of different data marks. The order of the tab pills can be reshuffled according to the presentation need. The indices are not an indication of the presentation order.

    Drag each index field into the Label shelf of respective data marks. Edit each Label which will be the text in the tab. Add mark, font, and color to format each tab indivisually.

    Create a Blank field and drag it to the Rows shelf. Turn off "Show header".

    3.Create story points 
    Each story point or slide will reside in a Floating container. There could be one or more sheets in one container. Please refer to this post on the creation process and how to use Layout Manager to precisely position the container. All containers are of the same dimension and placed on top of each other.

    In our example, there will be 7 story point containers, one per each point. Name each sheet with its index in the name, such as 1.Intro, 2.Bar Chart ...

    Note that in the example, we have turn on a couple of global filters. They stay visible through the entire deck.

    Using area annotation to create text is quite convenient.

    4.Create action filters
    We need one action filter for turning on/off each point. There will be 7 of them. (The other two are for navigation tab highlighting.)
     Use the tab sheet as the source and the story point sheets as target.
     The index fields will be used to target respective sheets.
    That's it. So this will give you more control over designing a deck of story points. You probably will spend more time in doing it than using the default story points. BTW the tabs can be vertical if you wish.
    Postscript
    Ville Tyrväinen has got some creative solution for the same issue. I have yet to understand his wizardry. He won't explain it because it may ruin further development. That certainly spurred my imagination. My method is somewhat different from his apparently. My post here might ruin somebody else's imagination. Very sorry about it. 
    3

    View comments

  3. Just wrote a post on back-to-back chart tweaks last week in my #TweakThursday series. This week's #MakeoverMonday project presents a dataset in which we can apply the technique. The idea is to provide back to back comparison at a lower level of details. This is to make up the shortcoming of the back-to-back chart, where the comparison at lower level is not obvious.

    So here you go
    When mouse hovering over the chart, you will see the detailed comparison at the gender level in the tooltips.
    One curious thing is, when published to the server, we see some horizontal lines in the solid color area. I already set the transparency to be 100%. Still the lines won't go away. Let me know if you know how to fix it.

    Since I try to follow the KISS rule (Keep It Simple and Stupid) this time, as promoted by Andy Cotgreave, I got time to create another simple version as follows. If you have a preference, I would love to know.
    Here is the one with visual tooltips.
    Click the above images to go to the interactive version.

    PS. I agree 100% with the KISS rule. In practice, we are required to turn out data analysis quickly, even though the look and feel is not graphically great. Some boring bar chart or line chart can save the day. So the most important requirement is turnaround time. Then, we can improve the graphics as the project goes. We are data analyst, not graphical designer after all.

    In the other hand, personally I like simple chart which is to make one point per chart. It will be easy to read and understand. Too much information cramped into one chart may be confusing.
    0

    Add a comment

  4. Back-to-back bar chart (also called divergent bar chart) is visually appealing and intriguing, I have to admit. Its shape is dynamic, capricious and unpredictable. It may look like a tree, a pyramid, a wave depending on the data set.

    However, a bit shortcoming of this chart is that, it is not obvious to compare any pair of the back-to-back bars. Overcoming this will make the back-to-back bar chart a viable solution in many cases.

    On 7/19/2016, the Viz of the Day or the original site is about population projection of a Swiss canton. It is based on a template from Prof. Dr. Ralf E. Ulrich's viz project. The prominent view is based on a back-to-back bar chart. It is very well designed: visually intriguing and full of information! The main problem I found is, the difference between man and women population is hard to see. And this difference is a very important piece of information.

    So, I made some tweaks to remedy this problem. Again, bars in tooltips come to the rescue.

    In summary, the tweaks I made are:
    - changed the reference line gray scale to make the man and woman halves discernible, without changing the overall tone of the chart. Left: New; Right: Original.
    - added bars in tooltips in the back-to-back bar chart to show the comparison between man and woman populations at every age.

    - added action filters to highlight the view at a given age, or at a age group.
    - added bars in tooltips in the summary table.

    With the above tweaks, we try to bring out the information that was not so apparent to the viewers. Hopefully, this will make this chart type more useful.

    Click images to go to the interactive version.
    0

    Add a comment

  5. Ben Jones of Tableau Software, the author of the book Communicating Data with Tableau, asked me if I can tweak his recent viz on the 50+ goals club of international soccer.
    It is quite an honor and a challenge! Ben's viz is already excellent. I would try to add some bells and whistles in my own taste.

    The main chart above is a scatter plot. Some time ago, Ben had a great article on creating scatter plot with lateral histograms. (It was called marginal histogram. Somehow, I like the word lateral better.) It is based on a design by Iron Viz Champion Shine Pulikathara.

    Scatter plot is a two dimensional way of calibrating the distribution of data. As a complement, it has been a custom for people to add lateral histograms to the scatter plot. They help us see the one-dimensional distributions, which can give us more insights along either of the dimensions.

    There are multiple ways to design the lateral distribution chart. Please go to Ben's article to see two variants. In the commentary, I mentioned a third with histogram +  box plot (see tutorial). Here I added it to the viz. Click the image to play with the interactive version.
    In summary, my tweaks are as follows:
    - Added lateral histograms + box plots on both Caps and International Goals.
    - Instead of using Size for [Goals per Match] which doesn't seem expressive, I created a lollipop chart for it.
    - Minimized some of the font size because the descriptive text is quite static.
    - Minimized the filters whose usage may not be high.
    - The key is to make the subject stand out.
    - Lowered the contrast a bit by using a gray background.
    - Added color to emphasize the variables in the tooltips.

    Postscript

    Recently scatter plot has been really hot. In the space of less than a month, 3 vizzies of the day are based on scatter plot. One of them is using Shine's design as template which includes lateral distribution charts. I would try to add lateral distribution charts to all of them.
    July 12:
    June 29:
    June 17:
    1

    View comments

  6. The Viz of the Day on 7/13/2016 is the first one from an author in China that I have ever seen. Although some of the description is in Chinese, the viz is understandable by anyone who knows English.
    The sensual curves of the sigmoid function and the wild cocktail colors make the viz irresistible to an ordinary man like me. The viz obviously seduced the committee of Tableau's Viz of the Day as much.

    Not familiar with this kind of chart type, I played with it a bit. And I found a few things to tweak.

    The viz showed successfully the ingredients in each of the well-known cocktails. But I also wanted to know the reverse: for an ingredient, how many cocktails contain it? All the data are available to answer this reciprocal question. I also needed to visualize the percentages of ingredients in a cocktail.

    So, initially this viz is about Cocktails > Ingredients discovery. My tweaks make it a viz of Cocktails <> Ingredients reciprocal relationship, based on the same data set.

    Here are my major tweaks:

    1.Ordering data 
    Putting order in the view is always a good practice for viewers to easily consume information. So I made all the following ordering from left to right.
    - Order cocktails by the number of ingredients each contains. That is, from the most sophisticated cocktail (Long Island Iced Tea), to the least one.
    - Order ingredients by the number of cocktails each is included in. That is, from the most used ingredient (Vodka) to the least one.

    2.Adding visualization 
    That shows how many cocktails one ingredient is contributing to. Four more action filters and two tooltip sheets are added.

    3.Use visual tooltips to list details
    The sigmoid chart is not the best chart type for the view. A horizontal bar chart can do a better job. Two tooltips are added: one for the cocktails and the other for the ingredients.
    Click the above images to view the interactive workbook.
    0

    Add a comment

  7. [First, this is mainly a server feature. You have to publish the viz to the server to see it in action. It doesn't work well on Tableau Desktop. There exists a curious discrepancy between desktop and server renderings of visual tooltips.]

    Here are a couple of facts about visual tooltips:
    • They are based on regular charts
    • They are turned on/off by action filters
    By laying "visual tooltips" on top of "visual tooltips", or "Viz on Viz" we can create recursive drilldowns.

    In theory, we can repeat the drilldown process infinitely. In practice, the process is limited by the number of dimensions in the dataset or by our imagination.

    An interesting feature of this technique is that we can do all the drilldowns in place/in context, instead of jumping to another tab or page.

    Let's see an example as follows, which is built on the Superstore dataset. Click the image to view the interactive version.
    There will be 4 levels of drilldown hierarchy on sales figures:
    • Year>Segment>Category>SubCategory
    The chart for each level is as follows:
    1. By Year: Vertical Bars
    2. By Segment: Vertical Bars + Title Sheet
    3. By Category: Donut Chart
    4. By SubCategory: Horizontal Bars + Title Sheet
    Feel free to pick the chart types you like in your own design.

    The main design steps are:
    - Create the above charts and sheets

    - Create a new dashboard.
    Here we use the sales by year bar chart as the base. Drag vertical containers (in FLOATING mode) into desired positions where visual tooltips will appear. Put every chart/sheet (in TILED mode) into a respective container. Remember to hide sheet titles. If you need titles, use another sheet. The reason is that the action filters won't hide regular sheet titles.

    - Create drilldown action filters between every 2 successive drilldown charts
    Note that each chart/sheet has to carry all the previous drilldown dimensions.

    - Use the layout manager at the lower left corner:
    1. Fine tune the position and size of each container. 
    2. Move lower hierarchy containers on top of higher hierarchy ones. Just press mouse and move.
    - Set the action filters to be "Select" and/or set the last drilldown action filter to be "Hover"

    - Validate all the drilldown filters in Desktop

    - Publish the design to the Server and validate the resulting viz again

    Voila, that's about it.

    A few further design tips:
    - Plan your visual tooltips first. The number of tooltips can be as many as that of data marks. We can reduce the number of tooltips by sharing one per multiple data marks.
    - We need one sheet per tooltip. Most of them are lookalike. We can create them by replicating.
    - Create a single hierarchy of tooltips and validate it first, before replicating tooltip sheets.
    - Create sets as filters that allow action filters to turn the sheets on/off.
    - Set tooltip background different from the overlapping chart.
    - Show successive dimensions in lower hierarchy tooltips.
    3

    View comments

  8. In 6/28/2016's Viz of the Day, it is charting the China Tea Leaf Index created by Reuters/Thomson, an index that is tracking the Chinese Economy. The index is composed of 10 factors. The formula is a nonlinear calculation. 10 factors are used as parameters that we can exclude one or more to see how much the index changes.


    When excluding some of the factors, however, we lost the view of the initial index with all the factors. So, the tweak I added is a pair of lines using dual axis. One line chart shows the original index with full factors. The other is with reduced factors. This way, we visualize the impact of some of the factors on the Tea Leaf Index. We see the difference if some of the factors are excluded.

    Here is the resulting viz by excluding the Tencent stock factor. We see that the index of the month is reduced by 17%. That's how Tencent can affect the index. However the impact is not linear in the calculation.


    One of the major objectives for data visualization is always to show the difference between different categories or scenarios, etc. The difference or the change is the information. Otherwise, we are not exerting the maximum impact that data visualization may bring about.

    The workbook can be downloaded here.
    2

    View comments

Blog Archive
Loading
Dynamic Views theme. Powered by Blogger. Report Abuse.