Sunday, 20 November 2016

Week 3 [21.11-27.11.2016] A Tour Through the Visualization Zoo

Hello,

I would like to present a topic of data visualization which I hope may be used in many of your research areas. Not only researchers have to deal with it. With a rapidly growing amount of data also companies became interested in data visualization. It should explain in a legible way what our data is about and help us to find required knowledge about some events. Due to a variety of data, this is not an easy task. Our data may contain time series, maps, networks or hierarchies. Each of these needs a different approach in visualization. An article presents some visualization techniques of various data. 


Additionally, you can watch short video, which presents some more interesting examples of data visualization in high dimensional space i.e. words represented by points with 200 dimensions
https://aiexperiments.withgoogle.com/visualizing-high-dimensional-space

1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?
2.  Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?
3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?

33 comments:

  1. Katarzyna thanks a lot for this interesting article. In my opinion this method to visualize data storage is very interesting and will be useful when it has been developed in the future. Thousand terabits of data should be visualized to be more flexible and available for everybody. The main goal of such kind of visualization should be to aid our understanding of data by leveraging the human visual system's highly tuned ability to see patterns, spot trends, and identify outliers. Well-designed visual representations can replace cognitive calculations with simple perceptual inferences and improve comprehension, memory, and decision making. By making data more accessible and appealing, visual representations may also help engage more diverse audiences in exploration and analysis. The challenge is to create effective and engaging visualizations that are appropriate to the data. Unfortunately this is my first contact with this topic but I hope not last.

    ReplyDelete
    Replies
    1. Thanks for your comment. I agree with you that visualization should be additional help to understand the data and a way to interest an audience.

      Delete
  2. 1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?

    That is super interesting subject. During Artificial Intelligence class there was a problem with showing data, but we only did it in 2 dimensional space. I have never tries to do it in high dimensional space. But after watching a video I see how changing 2D into 3D show data in much better way. I think that they did a great work and it is just a beginning of working on better visualisations of more demanding data. Not everything can be simplified to 2D. That seems to be a future of data presentation. We deal with virtual reality, where we want to see everything as in real life, look from different perspectives and 3D offers that.

    2.  Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?

    First thing that popped up in my mind was music visualisation. All the examples that I know are in 2D and they base of graphs, using different colours and all simple solutions. But when it comes to 3D I don’t know many examples. The only relevant example is Cascade (http://nytlabs.com/projects/cascade.html) used by New York Times to analyse data. It is quite old (have 5 years now) and Your example is much modern.

    3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?

    When it comes to people who have no idea about the subject I think that pretty data visualisation is much more helpful. They are looking for easy, understandable ways for complete newbies and scientific computations are not helpful at all. Taking myself as an example if I come up to a completely new topic, I do not start with searching of scientific data and complex scientific work.

    ReplyDelete
    Replies
    1. Cezary, thanks for your comment. I agree that sometimes it is much better to visualize in 3D than 2D. Fortunately, there are much more tools and techniques available.

      Your comment reminded me of this project: https://thomaslevine.com/!/user-2015/ . It shows a way to visualize data with music in R. There is an example on well-known Iris dataset and some other examples that look and sound weird but how knows it may be developed in the future.
      Project Cascada is very interesting and promising. Thanks for this example!

      Delete
  3. 1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?

    What interests me is possible applications of visualisation in higher dimensional space (beyond 3D). Some mathematical concepts might not be even properly imagined by us poor humans but for the use of e.g teaching/understanding Topology such tools would be extremely useful.

    2. Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?

    I personally love visualisation by Flowingdata (https://flowingdata.com/2015/12/15/a-day-in-the-life-of-americans/)
    which shows an average day in life of Americans. As the time of the day changes you can observe changing trends in people activities and how they move from one to the other. It is really mesmerising to watch and I am certain such visualisation has many possible applications.

    3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?

    As the saying goes, you need to have at least one image in your paper so that people would read it :) I think it is just our human nature that we absorb and remember information better when it is visualised.

    ReplyDelete
    Replies
    1. Ivona, thanks for your comment. I love this Flowingdata, you posted. It's absorbing that I have spent too much time staring at it.
      It's a good point that people need images not only to understand but also to remember information. It is important if we want our research not being forgotten.

      Delete
  4. 1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?

    I haven't tried visualizing, but I have conceptualized my approach to the problem. Then I realized that thinking about visualizing space of size over 32000 x 3 does not really make sense. Therefore I quit dreaming.

    2. Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?

    I guess for me scatter plot is still the top one :)

    3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?

    I don't think that eye-candy is the most important factor. If you really want to work with data rather than selling it to somebody, that is.
    As far knowledge inference is concerned, any technique that helps is good. It could be useful visualization, dimensionality reduction (i.e. Principal Components Analysis), anything. It doesn't matter. At the end of the day you need to gain something from data, otherwise there is no point to collect any datum at all.

    ReplyDelete
    Replies
    1. Paweł, thanks for your comment. I think you may try dimensionality reduction to visualize just the most important aspects of your data. Indeed, eye-candy is not the most important but when you work with data and want to show it to people then you need some good visualization.

      Delete
  5. I'm not a visualization expert. Frankly I'm really bad in any kind of UI/UX design. The only state-of-the-art visualization technique I know are ms excel charts and gnuplot. These simple tool do the job when it comes to show some statistic measures like central tendency or standard deviation to trim the data set or reject some features and therefore reduce the dimensionality.
    When it comes to visualizing high-dimensional data it's crucial reduce dimensionality and to pick some meaningful relation between two or three dimensions. You can then visualize data on 2D or 3D chart. Obviously you can repeat the step, choosing different features to present thus presenting data from different point of view. After all it all depends on who is the recipient.

    ReplyDelete
    Replies
    1. Mikołaj, thanks for comment. I've never thought of data visualization as a UI/UX problem. It's interesting point of view and actually it has many in common. Both are design to make science and technology easier.
      I agree that type of visualization should be customized to recipients.

      Delete
  6. 1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?

    The main primary goal of using data visualization is in education popularization and marketing. However I will not limit data visualization to 2D and 3D graphs. There is more potential in 3D animation. There are many considerable methods to visualize high dimensional space. Gnuplot is one of the tool that rightly know how to handle this kind of job.

    2. Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?

    There are so many techniques to visualize data that I realy cannot recall any of them.

    3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?

    At the beginning images are perfect tool not only to catch an attention but to improve understanding of subject. Before I started to analyze complex systems I started to learn Unified Modeling Language and this open me many doors of other modeling languages. Visualize data is another way to improve readability of data presentation.

    ReplyDelete
    Replies
    1. Piotr, thanks for comment. Yes, there are many methods to visualize more dimension than 3. We can add as a dimension colors even music, which I mentioned in one of the comments above.
      We can use visualizations also in different way than data, we can visualize for example concepts or UML.

      Delete
  7. 1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?

    Yes, once I tried to visualize harmonics of sound tones. It was a kind of high dimensional scatter plot. I used it to decode tones that were chosen as a phone number. I think that good data visualization is crucial in big data analysis.

    2. Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?

    I know only MATLAB – an environment for visualization (but not only). Moreover I heard about WolframAlpha but I didn’t use it for data visualization. In my situation it is hard to comment on it.

    3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?

    In my point of view the main aim of data visualization is to aid understanding of data by human visual system’s ability to see occurring phenomenon among data. I think that such visualization is really helpful technology. Scientific computations are not so clear and obvious.

    ReplyDelete
    Replies
    1. Damian, thanks for interesting example of data visualization. I've never had opportunity to visualize music but it sounds nice.
      I agree that scientific computations may not be clear without good visualization.

      Delete
  8. This comment has been removed by the author.

    ReplyDelete
  9. 1. Have you ever tried to visualise high dimensional space? What was that? If not do you know any area where good data visualisation is crucial?

    When Preparing my Masters Thesis I visualised financial data in order to find relationships between stocks and commodities. Modern financial data is high dimensional by nature. For example, one can monitor instrument prices in various timeframes. In the US the financial data is as detailed as milliseconds, which in terms of time series implies extremely deep analytics toolkit to be applied. Price and volume are the “plain vanilla” characteristics of instruments, but down to the fact that many factors determine the price of stock, analysts are forced to model many many inputs.

    2. Do you know any state-of-the-art examples of data visualisation techniques? Which one do you like most? Do you use or know some interesting visualisation tools?

    Due to my background, the most common data visualisation technique that I came across was time series analysis. In the presented article, authors from Stanford University define time-series data as set of values changing over time. The authors enumerate four techniques related to Time-Series Data, as follows:
    Index Charts;
    Stacked Graphs;
    Small Multiples;
    Horizon Graphs.

    For data visualisation, I use tools created by SAS, such as SAS Visual Statistics and SAS Visual Analytics.

    3. What is the aim of data visualisation? Do you think that people gain more knowledge from pretty data visualisation than scientific computations?

    In my opinion, the main aim of data visualisation is to present it in the most convenient and most valuable manner. It’s truism that a nicely presented dataset in form of graph or time series is by far adding more value to the recipient of the information that “pure” dataset itself. What is more by visualising the data in certain manner we may indicate specific phenomena or see trends which are not easily to observe in computations.

    ReplyDelete
    Replies
    1. Thanks for your input. Financial data is very intersting and prospering topic. Ability to visualize it is a skill that financial institutions are often looking for.

      Delete
    2. Thanks for your input. Financial data is very intersting and prospering topic. Ability to visualize it is a skill that financial institutions are often looking for.

      Delete
  10. Hello, thank you for an interesting topic. UX and good quality data visualisation is key in decision-making processes.

    1. Have you ever tried to imagine a high-dimensional space? What was that? If you do not know any place where good visualisation of data is key?

    Yes, I have. The results were spectacular. Unfortunately, the front-end is not my favourite part of the job :). As mentioned at the beginning of visualisation helps people look at the situation much faster than numbers.

    2. You know the state-of-the-art examples of data visualisation techniques? Which ones do you like most? Do you use or know some interesting visualisation tools?

    If we talk about the tools Matlab is the most popular in my opinion, but I focused on web solutions and I am currently using chart.js frame.

    3. What is the purpose of data visualisation? To Do you think that people gain more knowledge of data visualisation than pretty scientific calculations?

    The image is easier to interpret for people. We must remember that visualisation is an output of the calculation, and usually is converted to a specific write / pattern. Person who is involved in data analysis knows this patterns and it can easily find important deviations.

    ReplyDelete
    Replies
    1. Pawel, thanks for yout comment. I'm not sure if data visualization should be strictly compared with UX. It has something in common but it's a quite different task.

      Delete
  11. Unfortunately as a classic IT person I am not good at visualisation. However, while ago looking for an inspiration I found:
    "10 Best Data Visualization Projects of 2015" (http://flowingdata.com/2015/12/22/10-best-data-visualization-projects-of-2015/). On the mention site you can find some Tutorials and examples. (http://flowingdata.com/category/tutorials/)
    1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?

    Fortunately not. I will be real torture to watchers if I do so.


    2. Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?

    I do not know but my uncle Google knows everything.
    Please have a look at 50 Great Examples of Data Visualization. (http://www.webdesignerdepot.com/2009/06/50-great-examples-of-data-visualization/).

    3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?

    Visual information processing is the visual reasoning skill that enables us to process and interpret meaning from visual information that we gain through our eyesight. Visual perception plays a big role in our everyday life. It helps us in learning and interacting with others. Because of the ease with which we rely on perception, we tend to overlook the complexity behind it. Understanding how we interpret what we see can help us design and organize our visual information.
    For example Psychologist Richard Gregory believed that the visual perception relies on top-down processing.
    Top-down processing, also known as conceptual-driven processing, happens when we form our perceptions beginning with the big picture. We make our best guess of what we see based on expectations, beliefs, prior knowledge, and past experiences. In other words, we make calculated assumptions. According to Gregory, we are typically correct in those assumptions. (http://www.richardgregory.org/papers/knowl_illusion/knowledge-in-perception.htm)

    ReplyDelete
    Replies
    1. Marcin, thanks for your examples. There are really nice visualizations.
      Acctually, visual information processing is powerfull skill of humans and should be used in reaserch. That's why data visualization is so important.

      Delete
  12. 1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?

    Yes, I work with data every day. Currently I rarely create visualizations, but for few years I was doing mostly it. Visualizing high dimensional data usually boiled down to simplifying it with 2D graphics with additional properties as various colors and shapes, sometimes animations (to express for example timeline). In my opinion data visualization is important in many areas and it is difficult to pick the one that needs it the most.

    2. Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?

    I used MatLab and R language to visualize the data. Also the tools like Tableau or QlikView are easy to use for simple visualizations. Microsoft provides few add-ins to Excel for it. There are a lot of such tools on the market.

    3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?

    It depends. For example, the manager in a company usually needs nice and simple visualization of the data. The complex analysis will not be understood. From the other hand, the specialist in a certain area for sure will appreciate the visualization, but will be also interested in scientific computations and more detailed data.

    ReplyDelete
    Replies
    1. Emilia, thanks for your input. You are right that choosing between data charts and more datailed data depends on needs of project.

      Delete
  13. 1) Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?

    I've never tried to visualize high dimnsional space. I think that data visualization can be crucial there where a graph could be used. I imagine 3d data visualization as a graph. So, for example, it can be helpful in businesses like trade, shipment or logistics.

    2) Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?

    I don't know any state-of-the-art examples. This is not my field of interest and I'm very bad in visualizing anything. I used only RapidMiner on classes one year ago

    3) What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?

    As mentioned in the first article, the goal of the data visualization is to improve our understanding of data by using our abilities to recognize patterns and trends. I think that thanks to data visualization, we can make a decision much easier or have a general idea of what the data are saying much faster.

    ReplyDelete
  14. Hello,
    thank you for sharing this interesting article. I hope, that it will help me someday in my studies.
    Unfortunately, I have a little experience in high dimnsional space visualization. Mainly, I have experience with 2D. High dimnsional space visualization can be used anywhere: in time-series for stock moves, trade, services (OLAP cubes?) etc.
    I think I have a limited knowledge, however I think that tools like some of the R packages (ggplot2, RnavGraph?), Python packages, SAS, Matlab provide these capabilities. The main aim of data visualization is to acquire a better understanding of the data. We will never go through the numbers and anuderstand them easiliy. We can calculate statistical measures but it will never provide us with the proper information.

    ReplyDelete
    Replies
    1. Tahnks for your comment. I think OLAP cubes may be considered as interesting visualization of data but it is hard to understand by non-specialists.

      Delete
  15. 1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?

    No I did not. From my experience a good area for that could be ecommerce (big scale) where you have millions and sometimes billions (including relations) of data records that need to be analyzed to increase the performance of the business.

    2. Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?

    Unfortunately I don’t know any of them. I never had to use such tools

    3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?
    In my opinion such visualization is a helpful tool. When you go through a point where amount of text data is just too big to organize it properly and understand you need to go step further. I think that this is the moment where data visualization is able to help to understand the data in much better way

    ReplyDelete
  16. Thank you for a very interesting and forward-looking article. I have never worked with visualization of multidimensional data, but I think that this method allows you to notice increasing regularity. In my opinion the only multidimensional data analysis makes sense. Most traditional methods show "as is" often more interesting question is "why is this so?" Of course there are the traditional methods, which can examine the interdependence but they do not give such clear results. Methods of multidimensional data presentation help you better understand the different relationships and see some trends. Once we see these relationships we can explore what is the strength of this relationship but we need to know what to explore. As a society, we are at once more used to the transfer of knowledge through images because visuals are becoming more and more important.

    ReplyDelete
  17. 1. Have you ever tried to visualize high dimensional space? What was that? If not do you know any area where good data visualization is crucial?
    I've never tried and I'll probably never will. Morever, most visualization techniques can reduce usefulness of presented data. I prefer good old school tables :)
    2. Do you know any state-of-the-art examples of data visualization techniques? Which one do you like most? Do you use or know some interesting visualization tools?
    Probably MatLab and R language, however I've limited amount of knowledge in this matter.
    3. What is the aim of data visualization? Do you think that people gain more knowledge from pretty data visualization than scientific computations?
    They don't gain knowledge. Most of the time it's used for presentation and so on. It's looking good, but the added value is limitied.

    ReplyDelete
  18. Unfortunately, I didn't have a chance to explore this field of science. I can only provide my personal opinion with regards to usefulness of data visualisation. I think that data visualisation is a very interesting method that can help people understand more and get them interested in topics that they wouldn't even think of exploring do to their limited knowledge of particular field of science. Visual aids have always stimulated people's brains and have been more appealing than plain facts and calculations. That's how our brain works, we love pictures, we love images, we love sculptures, we love 3D. We love things that are engaging, entertaining ‒ things that improve our comprehension and memory.

    ReplyDelete
  19. Hey, first of all I would like to apologise for my late reply. Thank you for sharing this great article with us. Let me answer your questions:
    1. I have tried to visualize date but not on such level as described in the article and the video. I think it's very well suited in teaching.
    2. I am not really an expert on the topic however a quick Google search came up with these suggestions: http://www.creativebloq.com/design-tools/data-visualization-712402
    3. Data visualization may simplify complex subjects, therefore it might be used for introducing them to inexperienced individuals.

    ReplyDelete
  20. Many people are good eyesight, So the correct visual presentation helps in perception results. Sometimes graphic presentation may affect the receiver the same results positively or negatively. Eg. Scale manipulation. I do not use any special tools for data visualization, beyond the usual charts.

    ReplyDelete