Difference between revisions of "Talk:Lesson01"

From Visual Analytics for Business Intelligence
Jump to navigation Jump to search
 
(6 intermediate revisions by 4 users not shown)
Line 7: Line 7:
 
== Demystifying Visual Analytics ==
 
== Demystifying Visual Analytics ==
 
In this paper “Demystifying Visual Analytics”, by Christian Chabot from Tableau Software, is focused on debunking the common “stereotypes”/misconceptions with regards to visual analytics. Visual analytics is widely understood to reduces the cognitive load required to for analysis. It is widely believed to be used on massive data volumes. In response to this misconception, the author rebuked that the use of visual analytics on a small data set is just as effective as that of enormous data volumes. The key argument was that the complexity of answering questions often rises faster with the dimensionality of data rather than the number of observation. Visual analytics is exceptional in handling multidimensional data, regardless of its size. Therefore, even with a small dataset, the use of visual analytics is just as effective. The author also mentioned that visual analytics tools typically contains new visual paradigms, which can be a major put-off for new users who have to spend the time to learn a new analysis tool, is held untrue. Rather, new visual analytics software does not necessarily mean new paradigms. In the case of Tableau, it focuses on proven visual paradigms. The notion of using visual analytics to uncover hidden insight is contested by the author as well.  Visual analytics is concerned with improving the process of analytical reasoning. Thus, the uncovering of insights is merely part of the journey, not the destination itself.<br/><br/>
 
In this paper “Demystifying Visual Analytics”, by Christian Chabot from Tableau Software, is focused on debunking the common “stereotypes”/misconceptions with regards to visual analytics. Visual analytics is widely understood to reduces the cognitive load required to for analysis. It is widely believed to be used on massive data volumes. In response to this misconception, the author rebuked that the use of visual analytics on a small data set is just as effective as that of enormous data volumes. The key argument was that the complexity of answering questions often rises faster with the dimensionality of data rather than the number of observation. Visual analytics is exceptional in handling multidimensional data, regardless of its size. Therefore, even with a small dataset, the use of visual analytics is just as effective. The author also mentioned that visual analytics tools typically contains new visual paradigms, which can be a major put-off for new users who have to spend the time to learn a new analysis tool, is held untrue. Rather, new visual analytics software does not necessarily mean new paradigms. In the case of Tableau, it focuses on proven visual paradigms. The notion of using visual analytics to uncover hidden insight is contested by the author as well.  Visual analytics is concerned with improving the process of analytical reasoning. Thus, the uncovering of insights is merely part of the journey, not the destination itself.<br/><br/>
The use of visual analytics is not restricted to just complex problems, but to simple questions as well. On the contrary, it is also claimed that the use of visual analytics is designed to answer questions, be it complex or simple. It is also put across that simple questions are answered much more quickly using visual analytics. However, in relation to this argument, it can also be argued that simple problem can be solved without the use of visual analytics technology. And in all practicality, it could be better off without visual analytics. As much as a “simple” problem can entail, this simple software/application might just suffice to solve everyday problems. They might even do a better job in solving simple problems as compared to using visual analytics. I believe that there are 2 main factors which hinder the use of visual analytics to solve “simple” problems. Firstly, it’s the cost of such visual analytics software. Would the cost of using such software, justify in relation to solving a simple problem, which can be replaced by other software that does a more “inferior” job in data visualisation, but still yielding the same result. Secondly, for questions be it simple or complex, to be answered more quickly depends on the analyst’s ability to interact with the software itself. A skilled analyst trained on the platform would undoubtedly yield better result compared to an amateur one, regardless of platform.<br/><br/>
+
The use of visual analytics is not restricted to just complex problems, but to simple questions as well. On the contrary, it is also claimed that the use of visual analytics is designed to answer questions, be it complex or simple. It was also put across that simple questions are answered much more quickly using visual analytics. However, in relation to this argument, it can also be argued that simple problem can be solved without the use of visual analytics technology. And in all practicality, it could be better off without visual analytics. As much as a “simple” problem can entail, this simple software/application might just suffice to solve everyday problems. They might even do a better job in solving simple problems as compared to using visual analytics. I believe that there are 2 main factors which hinder the use of visual analytics to solve “simple” problems. Firstly, it’s the cost of such visual analytics software. Would the cost of using such software, justify in relation to solving a simple problem, which can be replaced by other software that does a more “inferior” job in data visualisation, but still yielding the same result. Secondly, for questions be it simple or complex, to be answered more quickly depends on the analyst’s ability to interact with the software itself. A skilled analyst trained on the platform would undoubtedly yield better result compared to an amateur one, regardless of platform.<br/><br/>
 
In my opinion, rather than focusing on the notion that visual analytics tackles both a simple and complex problem, a better argument would be how visual interaction with data compliments humans’ biased visual ability (we understand information better through graphic presentations). It gives the user an unprecedented ability to break down the complex problem into simpler, solvable parts. Even simpler problems would be made simpler with these visualisations. It allows people perform such tasks quickly because they let people apply computing operations on the data. People can go in any direction with their thoughts without needing to look away from visual representations that their brains can quickly interpret. It would save analyst previous time and effort.
 
In my opinion, rather than focusing on the notion that visual analytics tackles both a simple and complex problem, a better argument would be how visual interaction with data compliments humans’ biased visual ability (we understand information better through graphic presentations). It gives the user an unprecedented ability to break down the complex problem into simpler, solvable parts. Even simpler problems would be made simpler with these visualisations. It allows people perform such tasks quickly because they let people apply computing operations on the data. People can go in any direction with their thoughts without needing to look away from visual representations that their brains can quickly interpret. It would save analyst previous time and effort.
 
With much argument put forth in the use of visual analytics, it is indisputable that visual analytics allows the analyst to uncover and understand hidden insights. However, the most critical aspect in the process of analytics is the skill of the analyst himself/herself. Only the analyst can bring out the full value of the visual analytics.  
 
With much argument put forth in the use of visual analytics, it is indisputable that visual analytics allows the analyst to uncover and understand hidden insights. However, the most critical aspect in the process of analytics is the skill of the analyst himself/herself. Only the analyst can bring out the full value of the visual analytics.  
 
<br/>--Tan Kee Hock
 
<br/>--Tan Kee Hock
 +
<br/>
 +
== TWB v.s TWBX in Tableau ==
 +
For users who are using Tableau for the first time, I feel that it might be a good practice to always save your tableau files as Tableau Packaged Workbook(TWBX) instead of Tableau Workbook(TWB), which is the default setting when you save your work in Tableau. <br/><br/>
 +
TWB files contain information about the sheets, dashboards and stories that you have created in tableau but they do not contain any data. Thus, if you would like to share your workbook with someone, you would have to send both the TWB files and the data source files. <br/><br/>
 +
On the other hand, TWBX files are like “zip files” which consist of the information about the sheets, dashboards and stories that you have done in tableau and a copy of the data source file. TWBX files are more intended for sharing and analysis can be performed without internet connections to your data since the data is already present in this packaged file.<br/><br/>
 +
In order to save your tableau files as TWBX files, go to save as>save as type> select *.twbx.
 +
<br/>--Lim Hui Ting
 +
<br/>
 +
 +
== A Tour though the Visualization Zoo ==
 +
 +
I was fascinated by the readings of the visualization zoo as it has enabled me to appreciate the beauty of visual analytics in various forms. Prior to that, I could only imagine presenting data in pie charts, bar graphs, lines or table formats. By creating effective and engaging visualizations that are appropriate for the data, we are able to comprehend the story and information given to us, and make better decisions for the organizations that we serve in. Perhaps one of the most surprising discovery I've found is the existence of Protovis in all the examples shown, which is an open source language for web-based data visualization.
 +
 +
For aggregate data with lots of categories, it is wise to separate them out in small multiples, showing how each category performs with time. This will also allow us to get a sense of the correlating patterns for the different categories i.e. the unemployed U.S. workers increasing across all industries from 2008-2010 in Figure 1C. Parallel coordinates is also a strong method for visualizing correlations between dimensions. The more overlapping the lines are, the stronger the correlations between variables.
 +
 +
Choroleth maps that are visualized in the form of geographical areas are very fascinating as it symbolizes how the particular region is performing with respect to the rest of the country. It performs the function similar to a heat map, which can allow users to make informed decisions on various issues surrounding the states.
 +
 +
- Alson Tan
 +
 +
== Exporting data from Tableau ==
 +
I encountered this problem when I wanted to edit some calculated fields created by tableau using excel but had no idea how to export those data. So here are some solutions I found out:
 +
===== 1. You can export the data source on a Windows computer in either of the following two formats: =====
 +
 +
* Data Source (.tds) - contains just the information you need to connect to the data sources such as data source type, location, and custom fields. If you connect to local file data sources (Excel, Access, text, extracts), the file path is stored in the data source file.
 +
*
 +
* Packaged Data Source (.tdsx) - contains all the information in the Data Source (.tds) file as well as any local file data sources (Excel, Access, text, and extracts). This file type is a single zipped file and is good for sharing a data source with people who may not have access to the original data that is stored locally on your computer
 +
 +
===== 2.Exporting the Data from a Tableau Data Extract (TDE) to a CSV or Excel File or Access Table =====
 +
 +
* Right click on the Datasource in the Tableau Work Book
 +
*  Go to the option View Data
 +
* Copy and paste it manually to Excel
 +
 +
-Yang Chengzhen

Latest revision as of 23:40, 23 October 2016

Visual Analysis for Everyone

The whitepaper, published by Tableau, provided me with a great insight on the field of visual analytics. The examples illustrated in the paper strongly demonstrated the need for a data analyst to have an interactive visual interface in order to perform their job well. Prior to the course, I have been using Microsoft Excel as a main tool for data presentation and analysis. I felt that it was one of a few de facto software for visualizing data quickly and easily. However, after reading this white paper, I have a clearer understanding of what visual analytics actually entails and having the flexibility for users to express their thoughts freely on a user interface is indeed a powerful tool in the field of visual analytics. In this case, Microsoft Excel does seem constrained and restrictive for the analyst to perform analysis on the tool itself. Of course, if an analyst knows the end result of his/her analysis, Microsoft Excel can be a useful tool to generate those charts for data presentation. However, during the process of analysis where one has to frequently change data and views based on different context, it does not seem to be as useful as other visual analytics tools out there (e.g. Tableau, Qlik Sense etc.) where users can interact with the data/views directly.

To resolve existing data visualisation and analysis problems, Tableau implemented 5 principles into their software. Together with VizQL, it allows flexibility for users to express and change visualizations to answer questions as they analyse on the fly. The use of visualization best practices further allows an analyst to display their data effectively. In my opinion, one limitation of having best practices and standard charts/graphs may constrain user’s creativity to think of newer and better ideas to illustrate their analysis results. Although best practices are followed by many and it is understandable by everyone, there could be better ways to represent data better. This could not be easily achieved by the tool itself and requires time and effort on the analyst to decipher ways to do so. Despite so, Tableau is undeniably a good tool for all visual analysts to begin with analysing and exploring data to discover new insights and create additional business value for their organizations.
--Gwendoline Tan Wan Xin

Demystifying Visual Analytics

In this paper “Demystifying Visual Analytics”, by Christian Chabot from Tableau Software, is focused on debunking the common “stereotypes”/misconceptions with regards to visual analytics. Visual analytics is widely understood to reduces the cognitive load required to for analysis. It is widely believed to be used on massive data volumes. In response to this misconception, the author rebuked that the use of visual analytics on a small data set is just as effective as that of enormous data volumes. The key argument was that the complexity of answering questions often rises faster with the dimensionality of data rather than the number of observation. Visual analytics is exceptional in handling multidimensional data, regardless of its size. Therefore, even with a small dataset, the use of visual analytics is just as effective. The author also mentioned that visual analytics tools typically contains new visual paradigms, which can be a major put-off for new users who have to spend the time to learn a new analysis tool, is held untrue. Rather, new visual analytics software does not necessarily mean new paradigms. In the case of Tableau, it focuses on proven visual paradigms. The notion of using visual analytics to uncover hidden insight is contested by the author as well. Visual analytics is concerned with improving the process of analytical reasoning. Thus, the uncovering of insights is merely part of the journey, not the destination itself.

The use of visual analytics is not restricted to just complex problems, but to simple questions as well. On the contrary, it is also claimed that the use of visual analytics is designed to answer questions, be it complex or simple. It was also put across that simple questions are answered much more quickly using visual analytics. However, in relation to this argument, it can also be argued that simple problem can be solved without the use of visual analytics technology. And in all practicality, it could be better off without visual analytics. As much as a “simple” problem can entail, this simple software/application might just suffice to solve everyday problems. They might even do a better job in solving simple problems as compared to using visual analytics. I believe that there are 2 main factors which hinder the use of visual analytics to solve “simple” problems. Firstly, it’s the cost of such visual analytics software. Would the cost of using such software, justify in relation to solving a simple problem, which can be replaced by other software that does a more “inferior” job in data visualisation, but still yielding the same result. Secondly, for questions be it simple or complex, to be answered more quickly depends on the analyst’s ability to interact with the software itself. A skilled analyst trained on the platform would undoubtedly yield better result compared to an amateur one, regardless of platform.

In my opinion, rather than focusing on the notion that visual analytics tackles both a simple and complex problem, a better argument would be how visual interaction with data compliments humans’ biased visual ability (we understand information better through graphic presentations). It gives the user an unprecedented ability to break down the complex problem into simpler, solvable parts. Even simpler problems would be made simpler with these visualisations. It allows people perform such tasks quickly because they let people apply computing operations on the data. People can go in any direction with their thoughts without needing to look away from visual representations that their brains can quickly interpret. It would save analyst previous time and effort. With much argument put forth in the use of visual analytics, it is indisputable that visual analytics allows the analyst to uncover and understand hidden insights. However, the most critical aspect in the process of analytics is the skill of the analyst himself/herself. Only the analyst can bring out the full value of the visual analytics.
--Tan Kee Hock

TWB v.s TWBX in Tableau

For users who are using Tableau for the first time, I feel that it might be a good practice to always save your tableau files as Tableau Packaged Workbook(TWBX) instead of Tableau Workbook(TWB), which is the default setting when you save your work in Tableau.

TWB files contain information about the sheets, dashboards and stories that you have created in tableau but they do not contain any data. Thus, if you would like to share your workbook with someone, you would have to send both the TWB files and the data source files.

On the other hand, TWBX files are like “zip files” which consist of the information about the sheets, dashboards and stories that you have done in tableau and a copy of the data source file. TWBX files are more intended for sharing and analysis can be performed without internet connections to your data since the data is already present in this packaged file.

In order to save your tableau files as TWBX files, go to save as>save as type> select *.twbx.
--Lim Hui Ting

A Tour though the Visualization Zoo

I was fascinated by the readings of the visualization zoo as it has enabled me to appreciate the beauty of visual analytics in various forms. Prior to that, I could only imagine presenting data in pie charts, bar graphs, lines or table formats. By creating effective and engaging visualizations that are appropriate for the data, we are able to comprehend the story and information given to us, and make better decisions for the organizations that we serve in. Perhaps one of the most surprising discovery I've found is the existence of Protovis in all the examples shown, which is an open source language for web-based data visualization.

For aggregate data with lots of categories, it is wise to separate them out in small multiples, showing how each category performs with time. This will also allow us to get a sense of the correlating patterns for the different categories i.e. the unemployed U.S. workers increasing across all industries from 2008-2010 in Figure 1C. Parallel coordinates is also a strong method for visualizing correlations between dimensions. The more overlapping the lines are, the stronger the correlations between variables.

Choroleth maps that are visualized in the form of geographical areas are very fascinating as it symbolizes how the particular region is performing with respect to the rest of the country. It performs the function similar to a heat map, which can allow users to make informed decisions on various issues surrounding the states.

- Alson Tan

Exporting data from Tableau

I encountered this problem when I wanted to edit some calculated fields created by tableau using excel but had no idea how to export those data. So here are some solutions I found out:

1. You can export the data source on a Windows computer in either of the following two formats:
  • Data Source (.tds) - contains just the information you need to connect to the data sources such as data source type, location, and custom fields. If you connect to local file data sources (Excel, Access, text, extracts), the file path is stored in the data source file.
  • Packaged Data Source (.tdsx) - contains all the information in the Data Source (.tds) file as well as any local file data sources (Excel, Access, text, and extracts). This file type is a single zipped file and is good for sharing a data source with people who may not have access to the original data that is stored locally on your computer
2.Exporting the Data from a Tableau Data Extract (TDE) to a CSV or Excel File or Access Table
  • Right click on the Datasource in the Tableau Work Book
  • Go to the option View Data
  • Copy and paste it manually to Excel

-Yang Chengzhen