Information Visualization - Science topic

Md. Faishal Rahaman Final paper submission date is September 30, 2024. For more detail about the conference, please visit the conference website: https://ais.cn/u/MvUV7n

Just a quick answer to data visualization. I do highly recommemd to learn Python and use matplotlib to visualize data.  There are already existing many libraries focussing on data mining, AI, and machine learning.

There are available many courses online, books, and python manuals.

Learning Python to work with data is really worth it. For starters, the best is to find a YouTube video on the topic you want to solve or some close one.

References:

[1] Joakim Sundnes: "Introduction to Scientific Programming with Python", Springer, Simula SpringerBriefs on Computing (2010) ISBN 978-3-030-50355-0 (Open Access https://doi.org/10.1007/978-3-030-50356-7) 

[2] H.P. Langtangen: A Primer on Scientific Programming with Python, Texts in Computational Science and Engineering 6, Springer, DOI 10.1007/978-3-662-49887-3

I consider it ethical to prepare information graphics for an intended audience.In addition, in your portfolio bring both versions.

Sorry for the time taken to respond. And thank you all for your answers. I really appreciate it. I was searching more in the direction Lucía-Pilar Cancelas-Ouviña Lucí-Pilar responded me, in fact I found somebody using this app for editing. I know this things are used for script writing intensively, but Documentary editors, that have to structure tons of material sometimes without a real script, use some self generated boards, inherited from script writing, and sometimes or this is my hypothesis the visual diagrams used need other concepts to considerate than the ones script writers use. David Atkinson you are rigth, sorry if i did not stated that I was talking about non-fiction / documentary editing where the scripting happens during the editing part. Also I do not know much about scriptwriting itself so I also appreciate Hussien Ali Noor

Try

abline(v=X)

in a new line following your drapery-function.

I'm not sure what you mean. On p.125 of "Automating the design of graphical presentations of relational information" in figure 15 it says, "Accuracy ranking of quantitative perceptual tasks. Higher tasks are accom- plished more accurately than lower tasks. Cleveland and McGill empirically verified the basic properties of this ranking." Perhaps you have an older copy of the paper and it was updated?

Dear Reuben Reyes thank you for your comment, I must say that even the first sentence was already super helpful. I started to look for available volume rendering software.

I especially liked 3D Slicer. Using it I was able to visualize my core.

A step in the right direction. For example, NASA shared the images of Mars (india made rover landing site) enabling someone from India coming out with more information. So,I believe it is possible to extract more from given data (Beneficial to all).

Thanks Harsh Panwar

Here includes a download of the software. It is no longer in development, but according to a people a few years ago above, it was working then. https://support.clarivate.com/ScientificandAcademicResearch/s/article/HistCite-No-longer-in-active-development-or-officially-supported?language=en_US

One way of visualizing networks is by partitioning them. This is typically

done by subsequently removing a vertex or an edge with big betweenness. Betweenness is a measure for how often a vertex or an edge lies on a shortest route between two (systematically or randomly) vertices.

Instead of employing a shortest path search such as Dijkstra, you could use an

ant colony algorithm. Ants go from one point to another, leaving an amount of pheromone on an edge. The ants choose leaving a vertex and entering an adge with a probability distribution given by the pheromones found on the edges that

are connected to that vertex. Pheromone load on an edge decays with time, and

ants crossing the edge replenish it. This way, important edges can be identified

and be removed for clustering purposes.

Another method of dealing with big graphs could be to compress it (loss-ful).

You could combine vertices that are strongly interconnected to a single one.

Literature:

Regards,

Joachim

There are many visualization toolkit and their use depend on the kind of work you do. To name few :

Plotly

DataHero

Tableau

Raw

Dygraphs

ZingChart

FusionCharts

D3.js

HighCharts

Google Charts

Crossfilter

PolyMaps

Hi,

Check out the HighCharts library, I use it all the time to visualise different data, it is very easy to use - http://www.highcharts.com/demo , for your purpose I think heatmap would be ok  - http://www.highcharts.com/demo/heatmap , or you can use one of the charts with drilldown option.

The Club of Rome essentially did what you are asking for in its 1972 publication, "The Limits to Growth", by modelling a number of different scenarios. The results only of these scenarios were shown in a series of graphs, but not the underlying structure of the models. However, the structure of these models and subsequent models can be viewed and run using dynamic simulation software. One free software package is Vensim PLE. When using Vensim PLE (and other similar software) a conceptual model is first developed graphically using symbols for stocks, flows, and linkages. The direction of influence (positive and negative feedbacks) are then added. Quantification of influence is established by using equations and the model is then validated by comparing simulations against historical data and making necessary adjustments to the model. The validated model can then be used to create different scenarios by using slide controls to change the values of various parameters in the model. I recommend John Sterman's book "Business Dynamics: Systems Thinking and Modeling for a Complex World".

Colin Ware: http://www.bookdepository.com/Information-Visualization-Colin-Ware/9780123814647?ref=grid-view

Tamara Munzner: http://www.bookdepository.com/Visualization-Analysis-and-Design-Tamar-Munzner/9781466508910?ref=grid-view

Ward, Grinstein, Keim: http://www.bookdepository.com/Interactive-Dat-Visualization-Matthew-O-Ward-Georges-G-Grinstein-Daniel-Keim/9781482257373?ref=grid-view

& of course, anything by Tufte:

(using booko - I'm not an amazon stooge :) )

Sandeep,

 If you interested in exploratory data analysis on multivariate Gaussian Probablity  to improve your accuracy level ,you may explore , Calculating the Mahalanobis Distances and Probability Densities along with Hotelling’sT2 Statistic .

Alternative approach you may go with SIMCA,k-NN,etc various classifier in favor of pattern recognition technique.

This two books help you lot.

Well-come

I have taught problem solving, and tell the students that expressing a problem visually on paper is often a very helpful step.  Some research showed that for groups who were given the same problem, the ones who were told to draw it first came to an answer much more quickly than those who were not.  (Can't give the reference now.)

One of the difficulties is that students no longer write or draw.  The physical act of carrying this out helps memory.  In addition if the teacher draws something on the blackboard and the students copy it, they see how to construct it and are less likely to make mistakes by putting lines in the wrong places.  I find students get standard engineering symbols wrong because they copy without understanding.  This is also relevant to languages such as Chinese and Japanese, where the order of constructing the word is important.

I would say the educators have in general forgotten this.

The question focuses on the meaning of visualization in people's way collecting information. Vision is a  sense having highest capacity for collecting information in five people's natural senses, hearing, smell, taste, touch, and vision. Vision has been keeping 70 - 80 % information collected by a person every day. Hence, people perceive and understand easy information implied in image, graph. Map as a graph represents visually objects and fields at locations. The visualization of locations and attributes of objects and fields on a map assists people to perceive easy and understand better the objects and fields at locations. This is why people like map and geovisualization.

About the role of vision in people's perception of information may be found in:

S. Few, "Tapping the Power of Visual Perception," Perceptual Edge, pp. 1-8, September 4, 2004.

Self-Organizing refers to the way in which it locates clusters with minimum user input.

The process can start  by reading a  image, and then  using multi-component features of the image to organize the image pixels in groups.Also  heuristic process which increase the number of feasible solutions  can be added in order to speed up the process of optimal segmentation

Have a look at the open source Modelio UML software with integrated XMI support:

I really do not know the characteristics of the decision process in the big data world, but in the area of health, understanding risk information is enhanced when a format based on numerical data is changed to a information format using pictures, especially in the people with “low-numeracy” abilities. (see  Garcia-Retamero, R., & Cokely, E.T. (2013). Communicating health risks with visual aids. Current Directions in Psychological Science, 22 (5), 392-399. DOI: 10.1177/0963721413491570 )

A few of the other answers have given you specific direction, but I'm not sure there's a good answer here without more context.  What crime types are we talking about?  Are we using legal definitions (i.e., violations of a particular statute), broad categories (e.g., "robbery"), behavior-driven definitions (e.g., commercial robbery of liquor stores)?  

For what purpose are we visualizing / predicting crime?  Are we visualizing / predicting crime to allocate police resources effectively?  If so, then time of day / day of week is crucial.  If not, perhaps it is not.  Are we visualizing / predicting to prevent crime?  If so, we need to use definitions of crime that are theoretically relevant so we can produce effective prevention plans.  Are we visualizing / predicting crime for the purpose of siting a new business, in an attempt to ensure the location is likely to attract the target clientele?  If so, then crime is just one of many visualizations / predictions to make.

Since you mentioned hot spots, I assume we're talking about maps.  What's the geographic area to be visualized?  I'd use a different technique for a nation-wide map than for a neighborhood map.  Who is the audience?  Are they trained in mapping and/or stats?  If not, then we have to ensure that the map is readable to the general public, which means we should avoid fancy methods.

Basically, your question is like asking, "So, what types of statistics are in use by criminologists today?"  The only correct answer is:  All of them, depending on the problem we're looking to solve. 

You might want to look at Heidi Lam's research:

Some work have been done on visual perception of distance, Gladwell we'll talks about the "picture problem", and Wolfram discusses some limits to our visualization capabilities as humans. Personally l'll look for inspiration from Edward Tuft's efforts when an existing method fails me-perfect combination between numbers and art.

The comparison is made on which basis ?  In your paper , pseudo color image gives best result but that result is compared with which image ? I am confused about it. 

The MS result of true color's saliency map is obtained after comparison with which image ? 

Hi,

I'm currently working on an application (COMMON Tool) which helps researchers to visualize complex data from study of interactions between people working together.

This application is still currently in a prototype stage and is therefore not currently available.

For more information, visit the web site that presents this tool:

Unfortunately, this site is in French, the English version coming soon.

I specified this application on the basis on several research:

Bakeman, R., & Quera, V. (1995). Log-linear approaches to lag-sequential analysis when consecutive codes may and cannot repeat. Psychological Bulletin, 118(2), 272-284.

Bakeman, R. (2000). Behavioral observation and coding. In H. T. Reis & C. M. Judd (Eds.), Handbook of research methods in social and personality psychology. (pp. 138-159). New York: Cambridge University Press.

Jeong, Allan (2001). Sequential Analysis of Group Interaction and Critical Thinking

in Online Threaded Discussions. The American Journal of Distance Education, 17(1), 25-43.

Rohlfing, K., Loehr, D., Duncan, S., Brown, A., Franklin, A., Kimbara, I., . . . Wellinghoff, S. (2006). Comparison of multimodal annotation tools: Workshop report. Gesprächsforschung, 7, 99-123. 

Bakeman, R., & Quera, V. (2011). Coding schemes and observational measurement. Sequential analysis and observational methods for the behavioral sciences (pp. 13-25): Cambridge University Press.

Best regards,

Aurore Defays

Dear Sean Haughian,

Thank you very much for your own answer as well as pdf as reference. It is a wonderful help. I got a problem here. My students have been studying vegetation in lowland flat area. I decided and suggested to use NMDS axis scores from their sample by species matrix. 

I wonder which method would be appropriate. All ordination methods have this type of problem.

Thank you.

Chitra

Regarding images, the mutual information means how similar information is available among them. Therefore, a good approach is to compare images according to the CBIR (Content based image retrieval) methods. These methods return a similarity index. Using this index you can determine the similarity of a given image to any image in a database.

But if you want to see how two given images are similar, you may use the MSSIM *mean structural similarity index measure) or SSIM measure.

Usually, you'll need to make a trade-off between inserting items into your data structure and just accessing items. Based on how often you are going to change your graph and how often you just read from it different data structures are more efficient. There is no one-size-fits-all solution to your problem. Without deep knowledge of your problem it is impossible to give the right answer.

10000 elements is not a large number of items for today's computers. Almost every algorithm might be fast enough for you. Only if you need very high performance does it make sense to use the kind of hacks that Jochen has suggested. Otherwise a regular data structure will be sufficient (and a lot easier to maintain). Sometimes it is not worth it to spend a lot of time tweaking a data structure that nobody is able to maintain. So, ask yourself the following: Will the project stay for many years (10 years+)? Will this be a data structure that is only written once and never touched again? If the answer is "yes" to both questions you can consider looking for the optimal solution, but you don't have to. Don't spend to much time for complicated code that nobody else will ever want to touch, but they might have to. Be kind to your future self ;-)

I suggest the following reading list.

[1] Erik Duval, "Attention Please! Learning Analytics for Visualization and Recommendation", Proceedings of 1st International Conference on Learning Analytics and Knowledge (LAK),  2011.

[2] Martin M. Olmos, Linda Corrin, "Learning analytics: a case study of the process of design of visualizations", Journal of Asynchronous Learning Networks, 16 (3), 39-49, 2012.

https://www.youtube.com/watch?v=T4xRPyId_DU, openCV

hope this helps

Hi Reiner,

I would say the best method would be Structure-from-Motion (SFM) based methods. Ofcourse, the challenges with flying platforms would be vibrations and video stability but techniques exist to deal with these to great extent. Visual SFM by default would generate relatively sparse point clouds but with surface fitting and strong priors based reconstruction you can improve the results greatly, atleast in my experience. I would recommend VisualSFM (http://ccwu.me/vsfm/) to be a good starting point to experiment with.

Tauseef

OK. It works fine. Thks !

We are also interested by this topic ...

@Nikola : especially for table/matrix, tree/treemap, network/graph representation & visualization.

@Jan : we plan to study VISO in order to see what could be match our needs and complete our RoadMap vision for our SEAMLESS project which is a Semantic Cartography framework

We already uses very simple representation ontologies and visualisation ontologies developed by ourself ... but we want to go to a more "standard" approach

Our current approach (both business and SEAMLESS) is described here : https://www.researchgate.net/publication/260453376_How_to_improve_PLM_approach_efficiency_based_on_knowledge_engineering_knowledge_management_and_semantic_web_technologies_domains

We also set up a Semantic Cartography LinkedIn Group order to discuss and share about tool and visualization methodology driven by ontologies : https://www.linkedin.com/groups?home=&gid=8101187&trk=anet_ug_hm

Please visit our research web site http://ncva.itn.liu.se and try the visualization tools for exploring large statistical data sets.

Marianne,

There is a lack of research in this area, although some does exist. In addition to the references suggested above, I would add Colin Ware's textbooks:

Information Visualization

Visual Thinking for Design

He has also published some other works that might be of interest to you. In my own opinion, there are no really comprehensive rules for visualization, since visual interpretation seems to be highly personalized. I typically try to show my visualization idea to many people, and if they all understand it, I find I am usually on the right track. If not, then I probably need to modify it.

Another trick that I find helpful is to use some sort of 'key' or 'cipher' to help orient the audience to the figure. This can be something like a written analogy, or a visual legend showing how to interpret the figure.

Hope that helps.

There are many hidden relations among digital resources even if these resources appear to be far in the repository under different categories. Obvious hidden relations are co-citation and bibliographic coupling. Another hidden relation is collaboriative filtering, where users like A may also like B.

Information visualization could serve as a effective interface to utilize these hidden relations and give user the maxium flexiblity to navigate and access resources in the digital repository.

.

by the way, locally linear embedding (lle) is a nice alternative to mds ; should be rather fast for embedding in 2D

.

"..... and each of the features used are depending on what type of data mining that will be use..." is incorrect. Creators of the KDD dataset did not generate the data keeping certain DM algos in mind. Each feature is just a piece of information and it is up to you what you want to do with them. If you are not doing it already, I would advice to do a good literature survey on the relevant topics.