1. For visualizations that represent the distribution of the dataset, what are the advantages and disadvantages pie chart has over bar chart? My personal preference when designing this type of visualization is to choose bar chart. Because it is much easier to compare the size of bars (the height of the bar) than compare the size of pies ( angel/ 2 * Pi * Sqr(radius), which is a more complex computation for humans ). 2. I want to add up another category for the user aspects of VA, which is accessibility. For example, there should be a different interface for the old people and color blinded people. 3. The section of spatial visualization reminds me of the famous london Tube Map, which falls between the natural layout and abstract layout. It is a schematic diagram. It maps the real London Underground topologically rather geographically, and has deeply influenced other network maps around the world.
For your first question, I think the advantage of pie chart for me is that the pie chart concludes all the elements into one specific area. All elements cross the same center spot, so maybe you can evaluate the percentage or quantity based on the degree of angles. Also, for pie chart, all factors we considerate can also compare together with the whole value.
1 On page 4, the author talks about variable types of datasets. Since the design of data storage mechanisms, the data items are kept as rows of data fields, the author classifies them into text or numeric types. The author also gives a classification of types: categorical, which includes binary, nominal, ordinal, and numerical, which includes interval-scaled, ratio-scaled. This is confusing to me: how binary, nominal and ordinal relate to each other based on this context?
2 The author introduces several visualization types in this article, using diagram to show the relationship of data. One of those is Daisy Map. Icke and Sklar use this to present scores of a student. Why they doing this? How this kind of map differentiates from normal chart of statistics of students? What unique information can we get from this map?
1. Obviously, I should've read this article first, as it provides some introductory foundation-laying for the visual analytics newcomer like me. In the very beginning, Icke addresses the DIKW pyramid as I'd hoped to do related to the Lam et. al. article. I'm surprised that this article explicitly addresses wisdom (is this the first time we've come across this since learning about DIKW?). What does this imply about the VA field? It seems to be going beyond even the grandiose claims of KM! Acronyms! I also want to read Edward Tufte's book.
2. On page two of this article, it is revealed that "the field of VA was initiated by the US Department of Homeland Security after the tragic events of September 11." To what extent is this field still led, controlled, dominated, and/or influenced by Homeland Security and military concerns?
3. Poorly formulated thoughts: this article's discussion of spatial visualization, especially geo-spatial mapping, pairs nicely with some of last week's readings on the same and similar topics. I'm especially interested in using this type of VA to "map" literature and historical information, including the etymology of toponyms.
1. In one of the optional readings, “Data Visualization for Human Perception” by Stephen Few, the author discusses the ineffectiveness of pie charts. He claims that a pie chart does not accurately represent results, the data is not easily comparable (in that some slices appear larger than others but actually are not), and the values aren’t easily ranked. However, Icke presents a pie chart in this article and doesn’t mention its ineffectiveness. Is this just a difference of opinion, or should pie charts no longer be considered effective data visualization? Does Icke present any other methods that are also ineffective?
2. In the beginning of his article Icke describes how many believe that we are “drowning in data”, but then states that “collecting data is only the beginning of a long journey to wisdom” (1). However, does this line of thinking really justify anything? Specifically regarding government surveillance, could officials just say they were “searching for wisdom”, and their actions be condoned? Could hackers present the same argument – “I wasn’t hacking. I was just searching for wisdom”?
3. In comparing this article to Kramer et al.’s about sonification, did Icke present any data visualization techniques that could be better represented or enhanced by adding sound? Personally, I didn’t really see any. In fact, the ones involving spatial recognition seemed to be effective only visually. Is this just because visualization and sonification are apples and oranges, or because visualization is a better way to represent data?
1. On pg. 4 of the article, the authors mention different variable types found in data sets. They mention that there are categorical and numerical variables, but what about equations or unusual variable types? If the data in the data sets doesn't fall into either of those categories, could it still be used? Does everything belong to one or the other?
2. Under section 3.1.5, they discuss data in data sets that create challenges for data visualization. One of those difficult qualities is missing value for certain variables. But what about 'null' fields for certain data? Is null still technically a value? What would you do about variables like that - just discard them and find something better?
3. It seems like there is a ton of work that goes into selecting a correct data set. But because there are so many specifications and criteria to follow, don't you just tailor make a data set that will yield results that are easy to predict? Do people ever create data sets that haven't been defined and reduced down? Wouldn't that provide you with more accurate results for whatever system you are testing?
Section 3.1.2 on Dimensionality speaks about reducing the number of dimensions for visualizations in an effort to help analysis. Are there datasets or areas of research that would benefit from constant 3 dimensional visualization or benefit from a traditional 2 dimensional representation into a 3 dimensional one?
Icke notes that the first step of an analyst is to look at the data and see what “story the dataset tells”(4). Does this present any sort of a problem for things like “big data” and trying to find something meaningful amidst the noise? It seems like the end goal would be to find the story as it unfolds but with the amount of data available for analysis it would be easy to craft your own story rather than letting the data tell it.
The article begins with the hierarchy we’ve come to know very closely in the DIKW form. In an effort to reach that pinnacle, visualization could very well be a key component. However, in section 4 one of the problems acknowledge by Icke is the skill level of a user. How can we use visuals to their full potential if there’s an element of the DIKW hierarchy missing for some users hoping to understand the things presented to them?
1. On the first page of the article, Icke writes that the 1800s are the “beginning of the modern data graphics.” What does he mean by this? What makes visualization techniques like map-making so different in the 1800s from earlier times? Other than maps, where there other forms of visual analytics that were popular in past eras?
2. Facebook timeline seems to me to be a type of visual analysis as it displays my social interactions through the site temporally. What other types of visualizations could be used to analyze my data in Facebook? What kinds of information or patterns could we discover if we visually analyzed timeline information in other ways?
3. Icke spends most of the paper focusing on the system aspects of data visualizing, and a very small amount of space on user aspects and human machine collaboration aspects. Are there areas that he should have elaborated on for the last two aspects?
1. Page 2 talks about human-machine collaboration. “On the left, lies the fields which depend on humans exploring the data via visualizations, and on the far right are the field which more and more depend on automated analysis of data via mathematical and statistical methods and out little emphasis on human interaction and visualization.” How have certain fields been decided on which need more or less human interaction or computer interaction? Does it have mainly to do with how large the datasets are?
2. Under heading 3.1.5, Source and Quality, it lists three issues in dealing with data. Those issues are multiple data sources, uncertainty, and missing values? Are there any other issues that should be added to this list?
3. Under section 3.2, it says that “the ultimate goal is to analyze today’s large and high dimensional datasets as automatic as possible, since it is not practical for humans to visually inspect such complex datasets.” So how it this data inspected and analyzed if it is not looked at by humans? How is accuracy or correctness maintained in datasets if not checked fully by humans?
1. The ER Diagram is best used in domains where the entities and relationships are pre defined. In domains where there are huge chunks of data, can we analyze the relationship and keep building stages in the ER diagram as each relationship is determined? Is such an approach optimal?
2. In deriving a feature vector for a data item, what are the key requirements? What relationships need to be known between the data item and the corresponding element in the vector?
3. What are the limitations of temporal and spatial visualization? When do they fail to deliver results? In temporal visualization, if a layout changes every second, how do we choose the median values?
1. Icke makes note of the fact that interactivity between human and data is key in visualization. How has the ability to interact with data has changed the ways in which data is used, both in and outside of the realm of visualization? 2. Although visualizations often to not include language--what does the relationship between these two elements look like? Can something exist and be represented outside of a language construct? 3. Seeing the Chernoff Faces in Figure 17 made me think of emojis, Are emojis and their use in (texts, emails, etc.) really a kind of data visualization?
1 - I'm curious if visualization is useful only for large sets of data, or if visualization as a whole is one of the more useful systems to represent information. How does visualization offer accurate/clear data? For example, I could look at a pie chart, or an infographic detailing information, but how will I know if what I'm seeing is of statistical significance without also viewing the raw data?
2 - Can visualization oversimplify complex relationships? In thinking about issues like nutrition, or health care, or other multifaceted problems, how does one create a visualization that acknowledges the complexity of a situation while being aesthetically simple enough to convey a clear idea.
3 - If the "burden of interpretation," is on the user, what are the ethical implications of visualization? Can one just create a pretty infographic, a la the work of Information is Beautiful, then throw up their hands and say "well, it's your job to understand it all" when their data is misused or misinterpreted? It seems like visualization could easily be abused to become a new form of propaganda/misinformation if used improperly.
1. This reading reminded me of a recent project called ShotSpotter which helps cities and local police make visual maps of gunshots using acoustic microphones atop city buildings. If you are interested here is an article: http://www.washingtonpost.com/investigations/shotspotter-detection-system-documents-39000-shooting-incidents-in-the-district/2013/11/02/055f8e9c-2ab1-11e3-8ade-a1f23cda135e_story.html. It is interesting to see new patterns emerging which can give context and reasoning behind seemingly random events and this story illustrates that quite well.
3. When choosing visualization types to describe a particular dataset how do users know which model will prove most helpful? While I understand that some models would not be useful or practical there seems to be many ways that you could use models to interpret the same data.
2. The article discussed the interaction between data, computers, and humans, and proudly gives us the motto, "let everyone do what he does best". While this makes sense to me I'm also wondering what types of information might be lost in the process of parsing the decisions of computation out to the computer. Do we miss important information this way?
Very cool link, Graham. This is such a unique, useful way of instantly visualizing gun violence. This also reminds me of the article we read on linking text with geography in order to locate toponyms and make new scholarly connections.
1. On the first page, the author writes that “enormous amounts of data are populating storage devices and waiting to be transformed into some sort of useful information and then knowledge that would hopefully serve a good purpose” and then goes on to write that “unfortunately, the technologies to transform data into information and knowledge are far behind the technologies that collect and store data”. I’m assuming that visualization is the point at which data becomes information, but is it still considered data even within the visualized graph or chart or what have you, which only becomes information once fully interpreted and comprehended as a whole?
2. The author talks about the concept of dimensionality and how it poses problems for visualization and for analysis, and at one point seems to define data dimensionality as number of variables. He goes on to say that “the higher the dimensions get, the more the number of data items one needs in order to perform meaningful analysis”. I don’t really understand what he means by all this.
3. The author writes about the “the choice of visualization methods in order to perform a certain analytical task is generally the duty of the user”. It seems that the task very much informs the visualization, and I’m assuming that, depending on the data set, the visual representation will have to be more or less complex, right? I guess I’m just thinking in terms of pie charts or bar charts and their seeming simplicity, and how both of these methods seem rather straightforward as far as analysis goes, whereas Chernoff Faces maybe involves a bit more expertise and is not a method readily recognizable by most people.
1. I am confused with Figure 4, which shows that understanding principles will turn knowledge into wisdom. Ignoring the debates on the relationship between knowledge and wisdom, I just wonder can Visual Analytics realize the process from data to wisdom. Although VA could help human making decisions or even making decision automatically, I still think it has not reached the level of wisdom.
2. The author said because Visual Analytics is a human-machine collaboration in decision making, so it has three main dimensions. Every dimension can be separated into several parts. I wonder what is the ground theory of this demonstration. Why does Visual Analysis have three dimensions? Are there any scholars or relevant articles could give the reasons?
3. About the relationships between observations and relationships between variables, I get little confused. On the one hand, I wonder is there any relationship between observations and variables. I think that the independent variables might cause the observations. On the other hand, I believe the author’s opinion, “variables that have no relationship are called independent variable, is unreliable. Independent variables are related to the dependent variables and influence them. In my opinion, the independent variables are variables that cannot be changed by other variables.
1) I was fairly shocked when I read that, "the field of VA was initiated by the US Department of Homeland Security after the tragic events of September 11." Is the Department of Homeland Security still the primary research and development base of VA? What are some specific ways that they influenced its development?
2) Does the process of visualization clean up and simplify data? It seems like visualization can take complex situations and skew our perceptions of data.
2) I'm also interested in the potential of visualization to skew perceptions rather than facilitating them. Political maps, for instance, can display information in a manner that is technically faithful to the data but misleads the viewer as to its complexities (like all those electoral maps that show Texas as a big chunk of solid red and California as a chunk of solid blue). It's a topic worth exploring further, I think.
1) I had never heard of the Chernoff faces before, and I’m honestly a little baffled as to how they could be used to effectively convey data. The variations in the example given are often extremely subtle, and the relationship between the variations and the scientific variables being displayed is not clear. Humans do generally have a hard-wired ability to recognize human faces, but does that really mean that data pictured as a humanoid face is going to be easier to understand?
2) The examples of different data visualizations given by the article may have made more sense if the author had briefly noted which variables and relationships were being displayed in each image. Figure 19, for example, is an interesting data visualization, but it is not clear how the star glyph method is being applied here, or how it could be applied to other data. But this may raise an issue with visual analytics in general—how can we draw a distinction between data visualizations that make intuitive sense, and visualizations that are too abstract or obscure?
1. It lists several data analysis technologies at the beginning of the article, such as Exploratory Data Analysis, Information Visualization, and Visual Data Mining and so on. According to Figure 4, we use EDA for transformation from data to Information and visual analytics for transformation from knowledge to wisdom. Why can’t we use automatic technologies on the early stages of understanding?
2. It seems that there are overlaps between properties of datasets. So which should be the primary feature we could use to distinguish them?
3. It mentions that “Interactivity is the central concept in VA and through interactivity humans are allowed to communicate with the computer to provide feedback”. I think Human computer interaction has similar description, so what is the relation between HCI and VA?
1. Can visualizations really lead to wisdom? Is predicting possible future outcomes based upon post data analysis wisdom or knowledge? Are there other scenarios that one can do (like predicting future outcomes) from visualizations that are considered wisdom?
2. When sampling data from a larger data set to analysis and finding the correct visualization, what are some issues when one scales up? There is density of the visualization - is it becoming too much to see, understand or interpret? Are there other issues?
3. Are there more issues than multiple data sources, uncertainty and missing values when analyzing data? I would think format (file type and database) would be another issue. Are there more?
1. Visual Data Mining seems like an interesting topic and useful tool, in a similar way to sonification. Presenting large swaths of data visually can permit more rapid recognition of patterns than a large spreadsheet or report. Dr. Baldridge demonstrated some of this with his word mapping. What other applications might it readily apply to?
2. Creating visualization of complex data that isn’t cluttered or overly complex can be challenging. Often data-gatherers are not necessarily good visual designers. Is this something that could be a specialty area of graphic design, or is it already?
3. The facial visualization method is amusing, though the challenges listed in using our facial recognition remain steep, at least for data that is not easily classified with an emotional representation. This seems like it might be a good area for further research, both in our pattern-recognizing tendencies and how to pair the easily-recognized data of the face with analytics.
1. Before reading this article I had never seen or heard about Chernoff Faces, Daisy Maps, or Star Glyphs! While their designs are interesting, I have to wonder how well they actually convey information to the average person. Could this be a reason why they aren't seen or perhaps used as often? A pie chart may be boring, but for most people it tends to get the job done.
2. In section 3.7, the author states that 'visualization is the way data explains itself to the user'. But then in 5.1, there is the statement that 'the burden of making sense of the data is on the user'. However, given the extremely simplistic nature of some of these visualization systems, without viewing the data how can the user be expected to make complete sense of it? Especially since different systems can manipulate the context of the data just by simplifying it so much.
3. I have a friend who is extremely colorblind, to the point he has to memorize patterns on his shirt and keep everything in his closet in a set order, so that he can confidently pick out matching items to wear for work. How do visualization systems compensate for these kinds of situations, or conversely, does this build a stronger case for mixed visualization and sonification systems?
1. As mentioned on page 2, visual analytics is a highly interdisciplinary field, covering a wide range of fields from data management, data mining, perception and cognition, HCI, and visualization. So, what is the role of information science in visual analytics? What's the relationship between VA and IS?
2. What factors can influence the effectiveness of conveying information through visual analytics?
3. When discussing the user aspects, the author claimed that "the user has the final say on the decision to be made based on the data." What do "user" and "decision" mean here? Was he talking about people who use the VA system and decide which VA graph to use, or those who need visual analytic results and then take actions based on the VA graphs?
1. The author claims that dimensionality poses a problem for visualization because humans can only perceive three dimensions. What would be the purpose of representing more than three dimensions if humans would be incapable of perception beyond the third. Please explain dimensionality in further detail.
2. The Chernoff faces make an intriguing notion in data picturing, but how could they be used in a practical application beyond just understanding various combinations of facial characteristics?
3. In the conclusion the author tells us that the "burden of making sense of data is on the user." Wouldn't that burden fall on the designer of the system, the person who choses the type of visualization, etc.?
1. In this article the author uses a diagram to describe how the field of visual analytics is applied to the DIKW hierarchy. This diagram seems to indicate that visual analytics takes data and then skips knowledge and information to create wisdom. Do you agree that visual analytics is a method of creating wisdom directly out of data? What would be some of the drawbacks of bypassing information and knowledge? 2. In this article the author separates visual analytics into three main groups of aspects. These groups are system aspects, user aspects, and human-computer interaction aspects. Which of these three aspects do you think is the most important to visual analytics and why? 3. In describing the three aspects that were mentioned in the previous question the author spends more time discussing the system aspects than he does the user and HCI aspects. It seems as though there has been less work done on the user and HCI side of visual analytics work. Why do you think that there has been less work done on the user and HCI aspects of visual analytics? How does this compare to the distribution of evaluations we saw in the Lam et al. article?
1. "Wisdom is the ultimate state of having the understanding of the principles of a system that is being observed." I would argue that the "ultimate state" that the author is referring to here is the application of the wisdom gained from understanding observations. Gaining wisdom that doesn't allow for a practical application of it may not be completely useless, but if you have wisdom and don't use it for anything then why have it at all?
2. Is written language the most basic form of data visualization?
3. I'm not sure why this author included the small paragraph of 5.2. The author basically just applies the definition of utility to visual analytics systems. Is there something that I'm missing here?
1. ‘Visual Analytics (VA) is an emerging field that provides automated analysis of large and complex data sets via interactive visualization systems in an effort to facilitate fruitful decision making.’ I wonder how large and complex the data sets are. And why we need VA to do that? What are the limitations of VA?
2. ‘Visual analytics aims to cover all aspects of what the previous fields failed to cover, it is meant to be the whole process that provides means of visualization and analysis starting from the data till knowledge.’ Is that true? Can VA cover all the failed aspects? Is that definitely that VA can change data to knowledge?
3. When talking about the visualization types, the author mentions the bar chart and pie chart. So what is the difference between these two tools? Someone told me that the pie chart is the worst tool to describe data, why?
1. For visualizations that represent the distribution of the dataset, what are the advantages and disadvantages pie chart has over bar chart? My personal preference when designing this type of visualization is to choose bar chart. Because it is much easier to compare the size of bars (the height of the bar) than compare the size of pies ( angel/ 2 * Pi * Sqr(radius), which is a more complex computation for humans ).
ReplyDelete2. I want to add up another category for the user aspects of VA, which is accessibility. For example, there should be a different interface for the old people and color blinded people.
3. The section of spatial visualization reminds me of the famous london Tube Map, which falls between the natural layout and abstract layout. It is a schematic diagram. It maps the real London Underground topologically rather geographically, and has deeply influenced other network maps around the world.
For your first question, I think the advantage of pie chart for me is that the pie chart concludes all the elements into one specific area. All elements cross the same center spot, so maybe you can evaluate the percentage or quantity based on the degree of angles. Also, for pie chart, all factors we considerate can also compare together with the whole value.
Delete1 On page 4, the author talks about variable types of datasets. Since the design of data storage mechanisms, the data items are kept as rows of data fields, the author classifies them into text or numeric types. The author also gives a classification of types: categorical, which includes binary, nominal, ordinal, and numerical, which includes interval-scaled, ratio-scaled. This is confusing to me: how binary, nominal and ordinal relate to each other based on this context?
ReplyDelete2 The author introduces several visualization types in this article, using diagram to show the relationship of data. One of those is Daisy Map. Icke and Sklar use this to present scores of a student. Why they doing this? How this kind of map differentiates from normal chart of statistics of students? What unique information can we get from this map?
1. Obviously, I should've read this article first, as it provides some introductory foundation-laying for the visual analytics newcomer like me. In the very beginning, Icke addresses the DIKW pyramid as I'd hoped to do related to the Lam et. al. article. I'm surprised that this article explicitly addresses wisdom (is this the first time we've come across this since learning about DIKW?). What does this imply about the VA field? It seems to be going beyond even the grandiose claims of KM! Acronyms! I also want to read Edward Tufte's book.
ReplyDelete2. On page two of this article, it is revealed that "the field of VA was initiated by the US Department of Homeland Security after the tragic events of September 11." To what extent is this field still led, controlled, dominated, and/or influenced by Homeland Security and military concerns?
3. Poorly formulated thoughts: this article's discussion of spatial visualization, especially geo-spatial mapping, pairs nicely with some of last week's readings on the same and similar topics. I'm especially interested in using this type of VA to "map" literature and historical information, including the etymology of toponyms.
1. In one of the optional readings, “Data Visualization for Human Perception” by Stephen Few, the author discusses the ineffectiveness of pie charts. He claims that a pie chart does not accurately represent results, the data is not easily comparable (in that some slices appear larger than others but actually are not), and the values aren’t easily ranked. However, Icke presents a pie chart in this article and doesn’t mention its ineffectiveness. Is this just a difference of opinion, or should pie charts no longer be considered effective data visualization? Does Icke present any other methods that are also ineffective?
ReplyDelete2. In the beginning of his article Icke describes how many believe that we are “drowning in data”, but then states that “collecting data is only the beginning of a long journey to wisdom” (1). However, does this line of thinking really justify anything? Specifically regarding government surveillance, could officials just say they were “searching for wisdom”, and their actions be condoned? Could hackers present the same argument – “I wasn’t hacking. I was just searching for wisdom”?
3. In comparing this article to Kramer et al.’s about sonification, did Icke present any data visualization techniques that could be better represented or enhanced by adding sound? Personally, I didn’t really see any. In fact, the ones involving spatial recognition seemed to be effective only visually. Is this just because visualization and sonification are apples and oranges, or because visualization is a better way to represent data?
1. On pg. 4 of the article, the authors mention different variable types found in data sets. They mention that there are categorical and numerical variables, but what about equations or unusual variable types? If the data in the data sets doesn't fall into either of those categories, could it still be used? Does everything belong to one or the other?
ReplyDelete2. Under section 3.1.5, they discuss data in data sets that create challenges for data visualization. One of those difficult qualities is missing value for certain variables. But what about 'null' fields for certain data? Is null still technically a value? What would you do about variables like that - just discard them and find something better?
3. It seems like there is a ton of work that goes into selecting a correct data set. But because there are so many specifications and criteria to follow, don't you just tailor make a data set that will yield results that are easy to predict? Do people ever create data sets that haven't been defined and reduced down? Wouldn't that provide you with more accurate results for whatever system you are testing?
Section 3.1.2 on Dimensionality speaks about reducing the number of dimensions for visualizations in an effort to help analysis. Are there datasets or areas of research that would benefit from constant 3 dimensional visualization or benefit from a traditional 2 dimensional representation into a 3 dimensional one?
ReplyDeleteIcke notes that the first step of an analyst is to look at the data and see what “story the dataset tells”(4). Does this present any sort of a problem for things like “big data” and trying to find something meaningful amidst the noise? It seems like the end goal would be to find the story as it unfolds but with the amount of data available for analysis it would be easy to craft your own story rather than letting the data tell it.
The article begins with the hierarchy we’ve come to know very closely in the DIKW form. In an effort to reach that pinnacle, visualization could very well be a key component. However, in section 4 one of the problems acknowledge by Icke is the skill level of a user. How can we use visuals to their full potential if there’s an element of the DIKW hierarchy missing for some users hoping to understand the things presented to them?
1. On the first page of the article, Icke writes that the 1800s are the “beginning of the modern data graphics.” What does he mean by this? What makes visualization techniques like map-making so different in the 1800s from earlier times? Other than maps, where there other forms of visual analytics that were popular in past eras?
ReplyDelete2. Facebook timeline seems to me to be a type of visual analysis as it displays my social interactions through the site temporally. What other types of visualizations could be used to analyze my data in Facebook? What kinds of information or patterns could we discover if we visually analyzed timeline information in other ways?
3. Icke spends most of the paper focusing on the system aspects of data visualizing, and a very small amount of space on user aspects and human machine collaboration aspects. Are there areas that he should have elaborated on for the last two aspects?
1. Page 2 talks about human-machine collaboration. “On the left, lies the fields which depend on humans exploring the data via visualizations, and on the far right are the field which more and more depend on automated analysis of data via mathematical and statistical methods and out little emphasis on human interaction and visualization.” How have certain fields been decided on which need more or less human interaction or computer interaction? Does it have mainly to do with how large the datasets are?
ReplyDelete2. Under heading 3.1.5, Source and Quality, it lists three issues in dealing with data. Those issues are multiple data sources, uncertainty, and missing values? Are there any other issues that should be added to this list?
3. Under section 3.2, it says that “the ultimate goal is to analyze today’s large and high dimensional datasets as automatic as possible, since it is not practical for humans to visually inspect such complex datasets.” So how it this data inspected and analyzed if it is not looked at by humans? How is accuracy or correctness maintained in datasets if not checked fully by humans?
1. The ER Diagram is best used in domains where the entities and relationships are pre defined. In domains where there are huge chunks of data, can we analyze the relationship and keep building stages in the ER diagram as each relationship is determined? Is such an approach optimal?
ReplyDelete2. In deriving a feature vector for a data item, what are the key requirements? What relationships need to be known between the data item and the corresponding element in the vector?
3. What are the limitations of temporal and spatial visualization? When do they fail to deliver results? In temporal visualization, if a layout changes every second, how do we choose the median values?
1. Icke makes note of the fact that interactivity between human and data is key in visualization. How has the ability to interact with data has changed the ways in which data is used, both in and outside of the realm of visualization?
ReplyDelete2. Although visualizations often to not include language--what does the relationship between these two elements look like? Can something exist and be represented outside of a language construct?
3. Seeing the Chernoff Faces in Figure 17 made me think of emojis, Are emojis and their use in (texts, emails, etc.) really a kind of data visualization?
1 - I'm curious if visualization is useful only for large sets of data, or if visualization as a whole is one of the more useful systems to represent information. How does visualization offer accurate/clear data? For example, I could look at a pie chart, or an infographic detailing information, but how will I know if what I'm seeing is of statistical significance without also viewing the raw data?
ReplyDelete2 - Can visualization oversimplify complex relationships? In thinking about issues like nutrition, or health care, or other multifaceted problems, how does one create a visualization that acknowledges the complexity of a situation while being aesthetically simple enough to convey a clear idea.
3 - If the "burden of interpretation," is on the user, what are the ethical implications of visualization? Can one just create a pretty infographic, a la the work of Information is Beautiful, then throw up their hands and say "well, it's your job to understand it all" when their data is misused or misinterpreted? It seems like visualization could easily be abused to become a new form of propaganda/misinformation if used improperly.
1. This reading reminded me of a recent project called ShotSpotter which helps cities and local police make visual maps of gunshots using acoustic microphones atop city buildings. If you are interested here is an article: http://www.washingtonpost.com/investigations/shotspotter-detection-system-documents-39000-shooting-incidents-in-the-district/2013/11/02/055f8e9c-2ab1-11e3-8ade-a1f23cda135e_story.html. It is interesting to see new patterns emerging which can give context and reasoning behind seemingly random events and this story illustrates that quite well.
ReplyDelete3. When choosing visualization types to describe a particular dataset how do users know which model will prove most helpful? While I understand that some models would not be useful or practical there seems to be many ways that you could use models to interpret the same data.
2. The article discussed the interaction between data, computers, and humans, and proudly gives us the motto, "let everyone do what he does best". While this makes sense to me I'm also wondering what types of information might be lost in the process of parsing the decisions of computation out to the computer. Do we miss important information this way?
Very cool link, Graham. This is such a unique, useful way of instantly visualizing gun violence. This also reminds me of the article we read on linking text with geography in order to locate toponyms and make new scholarly connections.
Delete1. On the first page, the author writes that “enormous amounts of data are populating storage devices and waiting to be transformed into some sort of useful information and then knowledge that would hopefully serve a good purpose” and then goes on to write that “unfortunately, the technologies to transform data into information and knowledge are far behind the technologies that collect and store data”. I’m assuming that visualization is the point at which data becomes information, but is it still considered data even within the visualized graph or chart or what have you, which only becomes information once fully interpreted and comprehended as a whole?
ReplyDelete2. The author talks about the concept of dimensionality and how it poses problems for visualization and for analysis, and at one point seems to define data dimensionality as number of variables. He goes on to say that “the higher the dimensions get, the more the number of data items one needs in order to perform meaningful analysis”. I don’t really understand what he means by all this.
3. The author writes about the “the choice of visualization methods in order to perform a certain analytical task is generally the duty of the user”. It seems that the task very much informs the visualization, and I’m assuming that, depending on the data set, the visual representation will have to be more or less complex, right? I guess I’m just thinking in terms of pie charts or bar charts and their seeming simplicity, and how both of these methods seem rather straightforward as far as analysis goes, whereas Chernoff Faces maybe involves a bit more expertise and is not a method readily recognizable by most people.
1. I am confused with Figure 4, which shows that understanding principles will turn knowledge into wisdom. Ignoring the debates on the relationship between knowledge and wisdom, I just wonder can Visual Analytics realize the process from data to wisdom. Although VA could help human making decisions or even making decision automatically, I still think it has not reached the level of wisdom.
ReplyDelete2. The author said because Visual Analytics is a human-machine collaboration in decision making, so it has three main dimensions. Every dimension can be separated into several parts. I wonder what is the ground theory of this demonstration. Why does Visual Analysis have three dimensions? Are there any scholars or relevant articles could give the reasons?
3. About the relationships between observations and relationships between variables, I get little confused. On the one hand, I wonder is there any relationship between observations and variables. I think that the independent variables might cause the observations. On the other hand, I believe the author’s opinion, “variables that have no relationship are called independent variable, is unreliable. Independent variables are related to the dependent variables and influence them. In my opinion, the independent variables are variables that cannot be changed by other variables.
1) I was fairly shocked when I read that, "the field of VA was initiated by the US Department of Homeland Security after the tragic events of September 11." Is the Department of Homeland Security still the primary research and development base of VA? What are some specific ways that they influenced its development?
ReplyDelete2) Does the process of visualization clean up and simplify data? It seems like visualization can take complex situations and skew our perceptions of data.
2) I'm also interested in the potential of visualization to skew perceptions rather than facilitating them. Political maps, for instance, can display information in a manner that is technically faithful to the data but misleads the viewer as to its complexities (like all those electoral maps that show Texas as a big chunk of solid red and California as a chunk of solid blue). It's a topic worth exploring further, I think.
Delete1) I had never heard of the Chernoff faces before, and I’m honestly a little baffled as to how they could be used to effectively convey data. The variations in the example given are often extremely subtle, and the relationship between the variations and the scientific variables being displayed is not clear. Humans do generally have a hard-wired ability to recognize human faces, but does that really mean that data pictured as a humanoid face is going to be easier to understand?
ReplyDelete2) The examples of different data visualizations given by the article may have made more sense if the author had briefly noted which variables and relationships were being displayed in each image. Figure 19, for example, is an interesting data visualization, but it is not clear how the star glyph method is being applied here, or how it could be applied to other data. But this may raise an issue with visual analytics in general—how can we draw a distinction between data visualizations that make intuitive sense, and visualizations that are too abstract or obscure?
1. It lists several data analysis technologies at the beginning of the article, such as Exploratory Data Analysis, Information Visualization, and Visual Data Mining and so on. According to Figure 4, we use EDA for transformation from data to Information and visual analytics for transformation from knowledge to wisdom. Why can’t we use automatic technologies on the early stages of understanding?
ReplyDelete2. It seems that there are overlaps between properties of datasets. So which should be the primary feature we could use to distinguish them?
3. It mentions that “Interactivity is the central concept in VA and through interactivity humans are allowed to communicate with the computer to provide feedback”. I think Human computer interaction has similar description, so what is the relation between HCI and VA?
1. Can visualizations really lead to wisdom? Is predicting possible future outcomes based upon post data analysis wisdom or knowledge? Are there other scenarios that one can do (like predicting future outcomes) from visualizations that are considered wisdom?
ReplyDelete2. When sampling data from a larger data set to analysis and finding the correct visualization, what are some issues when one scales up? There is density of the visualization - is it becoming too much to see, understand or interpret? Are there other issues?
3. Are there more issues than multiple data sources, uncertainty and missing values when analyzing data? I would think format (file type and database) would be another issue. Are there more?
1. Visual Data Mining seems like an interesting topic and useful tool, in a similar way to sonification. Presenting large swaths of data visually can permit more rapid recognition of patterns than a large spreadsheet or report. Dr. Baldridge demonstrated some of this with his word mapping. What other applications might it readily apply to?
ReplyDelete2. Creating visualization of complex data that isn’t cluttered or overly complex can be challenging. Often data-gatherers are not necessarily good visual designers. Is this something that could be a specialty area of graphic design, or is it already?
3. The facial visualization method is amusing, though the challenges listed in using our facial recognition remain steep, at least for data that is not easily classified with an emotional representation. This seems like it might be a good area for further research, both in our pattern-recognizing tendencies and how to pair the easily-recognized data of the face with analytics.
1. Before reading this article I had never seen or heard about Chernoff Faces, Daisy Maps, or Star Glyphs! While their designs are interesting, I have to wonder how well they actually convey information to the average person. Could this be a reason why they aren't seen or perhaps used as often? A pie chart may be boring, but for most people it tends to get the job done.
ReplyDelete2. In section 3.7, the author states that 'visualization is the way data explains itself to the user'. But then in 5.1, there is the statement that 'the burden of making sense of the data is on the user'. However, given the extremely simplistic nature of some of these visualization systems, without viewing the data how can the user be expected to make complete sense of it? Especially since different systems can manipulate the context of the data just by simplifying it so much.
3. I have a friend who is extremely colorblind, to the point he has to memorize patterns on his shirt and keep everything in his closet in a set order, so that he can confidently pick out matching items to wear for work. How do visualization systems compensate for these kinds of situations, or conversely, does this build a stronger case for mixed visualization and sonification systems?
1. As mentioned on page 2, visual analytics is a highly interdisciplinary field, covering a wide range of fields from data management, data mining, perception and cognition, HCI, and visualization. So, what is the role of information science in visual analytics? What's the relationship between VA and IS?
ReplyDelete2. What factors can influence the effectiveness of conveying information through visual analytics?
3. When discussing the user aspects, the author claimed that "the user has the final say on the decision to be made based on the data." What do "user" and "decision" mean here? Was he talking about people who use the VA system and decide which VA graph to use, or those who need visual analytic results and then take actions based on the VA graphs?
1. The author claims that dimensionality poses a problem for visualization because humans can only perceive three dimensions. What would be the purpose of representing more than three dimensions if humans would be incapable of perception beyond the third. Please explain dimensionality in further detail.
ReplyDelete2. The Chernoff faces make an intriguing notion in data picturing, but how could they be used in a practical application beyond just understanding various combinations of facial characteristics?
3. In the conclusion the author tells us that the "burden of making sense of data is on the user." Wouldn't that burden fall on the designer of the system, the person who choses the type of visualization, etc.?
1. In this article the author uses a diagram to describe how the field of visual analytics is applied to the DIKW hierarchy. This diagram seems to indicate that visual analytics takes data and then skips knowledge and information to create wisdom. Do you agree that visual analytics is a method of creating wisdom directly out of data? What would be some of the drawbacks of bypassing information and knowledge?
ReplyDelete2. In this article the author separates visual analytics into three main groups of aspects. These groups are system aspects, user aspects, and human-computer interaction aspects. Which of these three aspects do you think is the most important to visual analytics and why?
3. In describing the three aspects that were mentioned in the previous question the author spends more time discussing the system aspects than he does the user and HCI aspects. It seems as though there has been less work done on the user and HCI side of visual analytics work. Why do you think that there has been less work done on the user and HCI aspects of visual analytics? How does this compare to the distribution of evaluations we saw in the Lam et al. article?
1. "Wisdom is the ultimate state of having the understanding of the principles of a system that is being observed." I would argue that the "ultimate state" that the author is referring to here is the application of the wisdom gained from understanding observations. Gaining wisdom that doesn't allow for a practical application of it may not be completely useless, but if you have wisdom and don't use it for anything then why have it at all?
ReplyDelete2. Is written language the most basic form of data visualization?
3. I'm not sure why this author included the small paragraph of 5.2. The author basically just applies the definition of utility to visual analytics systems. Is there something that I'm missing here?
1. ‘Visual Analytics (VA) is an emerging field that provides automated analysis of large and complex data sets via interactive visualization systems in an effort to facilitate fruitful decision making.’ I wonder how large and complex the data sets are. And why we need VA to do that? What are the limitations of VA?
ReplyDelete2. ‘Visual analytics aims to cover all aspects of what the previous fields failed to cover, it is meant to be the whole process that provides means of visualization and analysis starting from the data till knowledge.’ Is that true? Can VA cover all the failed aspects? Is that definitely that VA can change data to knowledge?
3. When talking about the visualization types, the author mentions the bar chart and pie chart. So what is the difference between these two tools? Someone told me that the pie chart is the worst tool to describe data, why?