Talk:Parallel coordinates

=== 2017 == Note for the editor

The "controversy" was resolved a while back SEE HISTORY in previous page so the polemics may be removed by the editor improving the overall article

2007

There is controvery over the history of parallel coordinates.

No there is not! The first IBM tech. report (140 p.) on the subject appeared in 1981 and first refereed paper (Visual Computer) in 1985 both by A. Inselberg. In 1990 E. Wegman published a paper in JASA where he acknowledged Inselberg's prior work. These are easily checked facts and not opinions (e.g. accessible via Google Scholar and directly). Informative articles should not fudge the facts nor contain self-serving statements -- "spin" -- which should appear(if at all) in the "Talk" section 132.67.20.213 14:05, 20 March 2007 (UTC)Ainselberg

Please keep in mind that others may have different opinion about the history to you ... It would be helpful if you cite the evidence/REAL references supporting\a different opinion than the above. Thanks ... 217.132.26.19 21:15, 11 March 2007 (UTC) SO WHY DON'T YOU CITE THEM? Italic text--- There are many articles and Tech Report by Ed. Wegman from 1980 which lead to his JASA paper that took 3 years to get published (of course you know how long JASA paper takes to get published)... Also, Inselberg totaly ignors Wegman's contribution and flips the fact every time. An example, there is a paper (chapter book) by Hinteberger cited Wegman's article on data visualization using ||-cord 1985. There are many articles we can find that are based on wegman's work either technical reports or published papers from 1980 till recently. I just like to give other the credits they deserve, that is all.

============

Wegman was at ONR in 1984 where he learned about ||-coords from A. Inselberg in the presence of B. Dimsdale. He then wrote a nice letter thanking me for the ||-coords break-through. There are no articles or Tech. Reports by Wegman on ||-coords prior to date. Wegman's "contribution" and where he lifted it from are documented below. It is worse than plagiarism for and his minions claim it as his own. AND NOW IT IS OFFICIAL -- see Edward Wegman: plagiarist deanesmay.com/2011/05/17/edward-wegman-plagiarist!Italic text89.138.89.13 16:55, 9 May 2007 (UTC) AInselberg

---

When making edits to this page, remember to sign your edits with four tildes, so that others can follow the conversation. Please refrain from making edits inline, and create a wikipedia account so that we can find out more about you and your affiliations. Hadleywickham 19:40, 10 March 2007 (UTC)

Regarding the history of parallel coordinates, early visualizations date back to 1883. Maybe there are even earlier ones. Please check http://www.davidrumsey.com/luna/servlet/detail/RUMSEY~8~1~32802~1152180 for a parallel coordinates visualization by Henry Gannett and F.W. Hewes, visualizing 10 dimensions with economic data for 47 US states. It´s been presented 1883. They did a lot more of these visualizations which can also be found in the archive. Regarding the question who invented this technique, this has seriously be taken into consideration as well ZH71 09:16, 17 February 2014 (CET).

General comments

This article isn't very informative at the moment. I'm currently researching Parallel Coords, and I would like to try to shape it up, but I'm worried that I will just be dragged into what seems to be a quarrel. Anyone involved in the writing still here? Merry 10:18, 22 January 2007 (UTC)

I'm not involved in the content of the article but agree that it isn't very useful at the moment. If you are willing to spend some time I would suggest a drastic prune to the supportable facts and then move the contraversial material to this talk page, where the discussion can continue. When/if there is a consensus then it can go back to the article. Rich257 10:43, 22 January 2007 (UTC)

Let us agree, this is not the place for arguments, "publications" or self-praise but for providing information on ||-coords based on published papers -- not in the proceedings of an obscure workshop.

Ainselberg 18:08, 12 May 2007 (UTC)

Controversial content

I have moved all this from the article as it needs further discussion and agreement before being added back. Rich257 17:35, 24 January 2007 (UTC)

The multidimensional system of Parallel Coordinates (abbr. ||-coords) was proposed in 1959 by Alfred Inselberg www.math.tau.ac.il/~aiisreal. The methodology's systematic development started in 1977 and shortly after it was applied to Exploratory Data Analysis (EDA) by J. Adams, A. Hurwitz, M. Boz, J. Helly, A.Inselberg at IBM-LASC and T. Chomut -- 1985 M.Sc. UCLA thesis on EDA. Much later Wegman 1990 introduced the PCP as a method for visualizing multivariate data after it had become a key tool in EDA and Data Mining (USA patent -- and more recently automatic classification). Other applications are on Collision Avoidance Algorithms for Air Traffic Control (1987 -- 3 USA patents), Linear Programming and Optimization, Intelligent Process Control, Computer Vision (USA patent), GIS, Business Intelligence, Financial Analysis, Intrusion and Fraud Detection and elsewhere by several researchers.

Parallel coordinates maps multivariate/multidimensional relations 1-1 into distinct subsets of the 2-D plane i.e. multimensional lines, curves, hyperplanes, hypersurfaces, families of "proximate" objects and their properties can be recognized from their planar images. There are distinct patterns for linearity, coplanarity, near-coplanarity, various singularities, proximities, convexity and non-convexity, non-orientable, developable, ruled surfaces and more. The textbook on Parallel Coordinates and their Applications will be released by Springer early in 2007. In the meantime the key papers starting in 1980, 1981(IBM reports), 1985(2), 1989 (Invited paper -- Amer. Statistical Soc. Annual Mtg), 1990, 1997, 1999, 2000 and more recently by A. Inselberg and others can be found from the WWW or by emailing aiisreal@math.tau.ac.il. Most of these papers are conferences and technical reports (this is FALSE) that describe the mathematics of the parallel coordinate based on projective geometry (also FALSE). The statistical properties of ||-coords and its impact in visualizing multivariate data can be found at E. Wegman (verifiably FALSE) 1990 and which were lifted directly from the first formula (duality) in the 1985 paper below. E. Wegman was in the audience at the Bureau of Census and elsewhere when applications of ||-coords to EDA were presented several years before 1990.

Recent algorithms for the Data Mining applications concentrate the relational information into clear patterns eliminating the clutter (heavy overplotting/overlapping) altogether. It turns out that relational information resides at the crossings of the polygonal lines. Replacing them by curves not only increases the computational complexity but more importantly distorts or destroys the information contained at the crossings. This of course depends on the interpolation function (see next paragraph) used to design the PCP. Moustafa and Wegman 2002 discussed the general frame work for designing the parallel coordinate plots using interpolation functions and show that the polygonal lines are constructed using the Piecewise Lagrange and the curves can be achieved using Splines. They are also proved that the PCP is an extension to Radon and Hough Transforms that existed many years before the proposal of PCP. The also described its relationship with Andrews Plot (Andrews 1972). More importantly, they proved that the lines/curves crossing is maintained using Splines and the computational complexity is negligible. In fact, the curves crossing in the PCP can be measure by computing the variance at this point, which is more preserved using Splines than the Lagrange. Moreover, the Energy norm (Parseval's Identity) is more preserved using the Splines than the Piecewise Lagrange, therefore, close points in the original space are close curves in the parallel coordinate space and hence it can be well discovered using data mining techniques, especially the density estimation and histograms techniques. It was H. Hinterberger in his 1987 Ph.D. thesis at ETH Switzerland who first studied and used density distributions in ||-coords. The original sophisticated application of ||-coords in Statistics in 1990 is by C. Gennings et al (Biometrics 46 719-735 -- a refereed paper) on Response Surfaces. In order to support your claims, please make these papers available!

Apropos the above remarks about conference papers the cited 2002 reference with no page numbers does not appear anywhere not even in conference proceedings. This paper is cited in more than article and its extension is refereed and will appear soon on Seeing the Million book. No PATTERNS based on "curved" ||-coords corresponding to multivariate relations have been published (known?) anywhere. There is a TREMENDOUS increase in computational complexity for finding intersections(crossings) between lines (O(N) complexity) and splines (O(N^6) complexity)! (FALSE...it means that the editor never looked at this paper. Even his background in computation and visualization is limitted. What is N? What is the computational complexity to compute the basis (1-t,t) for (t=[0,1]) and (cos^3(t),sin^3(t) for t=[0,pi/2])? Please, read the Moustafa and Wegman 2002 to confirm that the used Spline-like basis doesn't increase the computational complexity ). Further, curves may be close but their intersections (i.e. the crossings) are not necessarily close (This comment is written by someone with limited background in mathematics and statistics ...Simply compute the variance at the point of duality using Splines then using the Lagrange and perhaps add noise to the data to see its impact on the point of duality in TPCP and SPCP designes. The Splines is more bounded than the Lagrange and hence the point of duality will not be affected much by the additive noise compared to the Lagrange. We teach this fact to our undergraduate students), as in the vicinity of an axis where they are nearly parallel. In short, the crossings are terribly distorted destroying the relational information (FALSE). These are elementary observations (I hope this comment is made by statistician or someone understand the parameter transformation process ) and indicative of why this idea has not been accepted for publication for 4 years after it was proposed( We collected the evidence to prove that the referee on that paper at that time has little statistical background and more importantly he/she is not well aware of the literature and is not honest referee. Simply he mentioned several names that dealt with this problem and their emails confirmed his lies, hence the EIC directly removed his name from the referee's list). The essence of ||-coords is not how the values are joined but what PATTERNS can be discerned from the display. It is these patterns which reveal multivariate relations in a dataset from its ||-coods display. ----One simple pattern revealed using Splines is the clear quantization level on the ||-axes that best describes the data distribution, even for very large data sets. This pattern never depicted using the Traditional PCP. By the way, as an easy comparison can verify, the "statistical" results claimed in Wegman's 1990 paper were lifted (with errors!) from the first formula (duality) in the 1985 paper. ---This paper is reviewd and published in JASA-- so it is hard to believe that there is any overlap between Wegman's work and Insleberg. Thier approaches are different. --- Insleberg used the ||-coord to study well defined geometric objects and Wegman is to study multivariate data without knowing the underlying structure(s). The 1985 discussed the image of geometric objects in PCP and never discussed the corrleation coefficient or the detection of clusters as described in Wegman's 1990. The touted 1990 paper is the ONLY refereed article on ||-coords with Wegman's name. The up to date authoritative reference on the subject is the chapter on ||-coords in the 2007 Handbook on Data Visualization (Springer series on COMPUTATIONAL STATISTICS) selected and refereed by the experts on Data Visualization in the Statistical Community. Which means it will be the fourth refereed paper on the subject after the Wegman's 1990 paper, whcih is considered the landmark paper in visualizing multivariate data with PCP regardless of many false facts in the PCP history. Note,the Moustafa and Wegman 2002 is selected and refereed by the experts on Data Visualization in the Statistical Community and appeared in "Graphics of Large Datasets: Visualizing a Million (Statistics and Computing) pp. 143-155, edited by Antony Unwin, Martin Theus, Heike Hofmann, Springer 2007. Although I hold much respect to A. Inselberg for his major contribution to the Parallel coordinate plot, his work is based on the projective geometry that is completely different approach than the parameter transformation discussed in Moustafa and Wegman 2002, which allows the discovery of important relationships in PCP such as the preservation rate of the Energy norm, mean, variance, correlation, etc. These relationships never studied in any PCP-articles (refereed or non-refereed). The basis function that preserves the higher rate of information throughout the trnasformation is the Spline (used to design Smooth PCP) not the Lagrange (used to design the Traditional PCP) and the impact will be on the accuracy of pattern detection in the parameter space. This discussed in the paper that he/she rejected but other honest referees believe its novel idea, and siginificant contribution to PCP, and will appear soon.

Short History and Development of Parallel Coordinates

Note to the editor: Please review this portion to ensure that the contents are verified and not controversial.

The multidimensional system of parallel coordinates was proposed by Alfred Inselberg in 1959 (see www.math.tau.ac.il/~aiisreal). Since 1977 the concept and results were presented in several conferences and seminars. The first refereed article on parallel coordinates (abbr. ||-coords or PCS) appeared in 1985 (A. Inselberg , Visual Computer, 1 , 69-91 with the line : x2 = m x1 + b --> (1/1-m , b/1-m) point mapping (derived in 1959) on the first page. It is well known that complete dualities can ONLY exist in the Projective Plane. This is a mathematical consequence with important ramifications and not a personal preference. Here the line with slope m = 1 is mapped into the direction (ideal point) with slope b – something meaningless in the Euclidean Plane. Starting with this duality about 40 deeper results were derived and presented in the 1985 paper which were applied in several fields including Statistics (see Amer. Statistical. Assoc. 1989 annual meeting -- invited paper -- (Stat. Graphics, ASA 1989 Proc. ,1-16). In 1990 two important papers on the subject appeared, C. Gennings et al applying ||-coords for the construction of Response Surfaces (Biometrics, 46, 719-735) a truly sophisticated and ORIGINAL statistical application. The other, by A. Inselberg & B. Dimsdale in the Proc. IEEE Vis'90 361-378 introduced recursion extending the point <--> duality to coplanarity and near-coplanarity, is by far the most widely quoted paper on ||-coords.

It is the PATTERNS in a ||-coord display which contain relational information. There may be a zillion polygonal lines which if they ( nearly) cross show linear relationship. Then only the points need to be retained. With recursion (1990) this has been generalized to complex non-linear multivariate relations and automatic classification for Data Mining. The work has appeared in refereed papers in the JACM, SIAM and Computational Statistics journals, the critically refereed Proc. IEEE Vis conferences and elsewhere. All this is contained in the forthcoming textbook on Parallel Coordinates Springer 2007. A synopsis is the chapter on ||coords in the Handbook On Data Visualization (Springer – series on Computational STATISTICS), the up-to-date authoritative reference selected and refereed by the experts in Data Visualization within the Statistical Community.

With relational information concentrated via recursion at the crossings, "curved" parallel coords not only greatly increase the computational complexity (e.g. from solving -- for the intersection -- linear to higher power equations for curves) but also destroy the location of, and hence information contained in, the crossing points. For example, two nearly parallel curves may be close but their intersection is far. The minimum energy claim cited elsewhere is of interest in Elasticity Theory (i.e. for the bending of beams) but irrelevant here.

Parallel Coordinates is a METHODOLOGY (as reflected in the 500 page fortcoming textbook -- Springer). Calling it a "plot" (like pie-charts, histograms etc) simply reflects ignorance, consistent with some of the remarks and claims made elsewhere(below/above).

The list(chronologically) of important original work on ||-coords includes that of B.Dimsdale, J. Adams, M. Reif, T. Chomut, M. Boz, H. Hinterberger, J. Eickemeyer, M. Ward, A. Chatterjee, C. Gennings, C. K. Hung, G. Grinstein, D. Keim, C. Jones, A. Desai, T. Matskewitch, J. Dykes (GIS), N & G. Adrienko (GIS), S. & T. Avidan, N. Helfman, A. Goel, M. Ankerst, L. Yang, J. Johansson, Y. Yaari (Visualizing Complex valued functions), Y. Singer & Greenspan (Visualizing structure of large networks), A. Shaus(Visualizing derivatives). Please send additional nominations to aiisreal@math.tau.ac.il as well as additional information on the contents. 132.67.20.213 14:05, 20 March 2007 (UTC)Ainselberg

Anyone else please demonstrate decency, professional standards and ORIGINALITY by writing your comments SEPARATELY

The first refereed article on ||-coords appeared in 1985 from which in 1990 E. Wegman lifted his elementary "statistical observations" . This can can be verified by an easy comparison, the "statistical" results in this paper were lifted from the point -- line duality in the 1985 i.e. lines with negative slope are represented by points between the axes. Much before that time, the technique had already become an important tool for those who attempt exploratory data analysis and visual data mining. E. Wegman was in the audience when statistical applications of ||-coords were presented in the Census Bureau and elsewhere many years BEFORE 1990. The editor contradicts himself here because there are several technical reports and conference papers on the EDA were published by E. Wegman and his students at the eairly 1980 and the Wegman's 1990 ., appeared in JASA, is the first refereed journal paper in the history of parallel coordinate plot. This is an utter falsehood -- e.g. see 1985 paper below. Using the point <--> line duality in the 1985 paper it repeats with errors important statistical properties of the PCP, such as the visual detection of correlated data, clusters, and classified information. These properties were pointed out by Inselberg at the 1989 ASA annual meeting -- invited paper. Some other fundamental statistical properties of the plot such as the detection of the underlying data distributions (uniform and normal) are discussed by J. Miller and E. Wegman 1991. This is a leading article in dealing with very large data visualization (clutterd parallel coordinate plot) (like 5 variables and a few data entries( 1000 points can be enough to show the clutter effect in PCP and how density estimation introduced by H. Hinterberger in his Ph.D. thesis at ETH Zurich can deal with it)), which is recognized recently as visual data mining. By the way 1985 < 1990!!!.

I'm not sure that the editor had the chance to read that thesis. The work by Wegman and Millir established the frame work for estimating the density in Parallel Coordinate, which is the first and most important paper in mitigating the clutter effect in PCP and now is recognized as Visual Data Mining. I will leave this to the Editor to confirm to me that this was done by Hinterberger. Again, the editor here must read the articles first and make it aviablable.

Unfortunately, A. Inselberg used to search for any article with some keywords and believed that the any article has his keywords must address same topic. For example, he searched on density estimation and parallel coordinate plot, the Google shows link to Henterberger. Then without reading the article, A. Insleberg declares that Henterberger introduced density estimation method for parallel coordinate plot. This is outrage, simpley the referees and honest scientists must read the articles and give honest assessments in their review. My comment on Henterberger work is that he tried to compute the dense regions in the original data space then the centeriod of each dense region is visualized in parallel coordinate plot. Interestingly enough he was motivated by a 1985 technical report by Ed. Wegman for visualizing multivariate data using PCP. This also confirms that most of the work in multivariate visualization using parallel coordinate is based on Ed. Wemgan's work. I would like honest referee to explain how Henterberger method can be related to Miller and Wegman 1991 work in estimating the density in the parallel coordinate and replace the lines by their density.

Another comment, Insleberg mentioned that the SPCP existed before Moustafa and Wegman 2002 work. Why didn't he show the date and links to the article? More importantaly, the generalized parallel coordinate plot (GPCP) is developed in Moustafa's PhD dissertation, GMU, 2001. And the goal is not only smoothing but discussing the parameter transformation in visualizing multivariate data, of which the parallel cooridinate can be a special case.

References

Modern trend

Thus, in applications such as data mining of multivariate datasets, the modern trend is to concentrate relational information into clear patterns eliminating the polylines altogether. This idea STARTED with ||-coords i.e. lines in N-D map into N-1 points is the first such pattern.

<--Please verify that this idea is never discussed before, I still belive that this idea was first introduced by Miller and Wegman(1991)... See the related work by, Ward, Artero, Andreko, etc..Although they draw the ||-coordinate horizontally. [A Inselberg, I think] —The preceding unsigned comment was added by Hadleywickham (talk • contribs) 14:22, 12 March 2007 (UTC).

Checking Google Scholar, the Wegman & Miller 1991 paper has 13 citations and there are NO (i.e. zero) citations for Wegman's & Moustafa's work. This is the way professionals measure influence in the field. Still (the writer below) is not confused by facts! If people used the claimed publications how come they did not reference them? Do you know what the word CITATIONS means?

UPDATE -- Feb. 2012 Citations: Wegman 1990 paper (The "landmark") 494, Inselberg 1985 paper 715, Inselberg & Dimsdale 1990 paper 763 Italic text Ainselberg 18:14, 12 May 2007 (UTC)

217.132.201.9 22:23, 10 May 2007 (UTC) AInselberg

<---We know what citations mean. There are 3 citations for Wegman's & Moustafa's 2002 workshop paper (which is presented in front of you). Two of these citations are from your place and the third is by LANL students. You need to find these articles and read it or will send it for you. Our advise to you for the second time is not to relay only on Google.. Wegman's & Moustafa's paper is made available to public recently. It officially appeared late 2006 in the large data visualization book, and it will have great influence. It is a novel approach to the parallel coordinate plot. Simply it is a new school of thought to the plot. Regardless your comment, researchers will find out soon how this is very important article in order to study, explain and extend the theory of parallel coordinates.

By the way Wegman & Miller 1991 has 16 citations not 13, and Wegman 1990 has 183 citations not 176 as you claim. We can show you more citations for these papers that never been listed on Google.

==============

<---The lack of influence is not true argument because most of the work in visualizing large data is based on Miller and Wegman 1991 Article. Moustafa's work is a new vision of the traditional PCP. It is a significant contribution to the theory of PCP. Many problems that never been explored can be easily discussed on the light of parameter transformation (Moustafa and Wegman 2002,2006) than the projective geometric approach by (Inselberg 1985, Wegman 1990). Problems that are not well addressed and inherent in most softwares can be resolved if we really follow the parameter transformation approach. One example is the enveloping method (which has so many drawbacks) can be well studied in the view of parameter transformation.

I've removed references to the "Generalized Parallel Coordinates Plot", since it's not a generalization of parallel coordinates geometry at all, but rather a variant with splined polylines. The term "generalized parallel coordinates" does not appear to be in common use in papers, articles or software on parallel coordinates (excluding the original paper). I've kept a single paragraph on splines better revealing quantization, the "smooth parallel coordinates plot" and its relation to Andrew's Plot. Ohbias (talk) 09:44, 5 September 2012 (UTC)

By contrast, the Wegman 1990 paper has 183 citations, Inselberg's 1985 paper 192 and the Inselberg & 

Dimsdale 1990 paper has 297. One of Ward's major contributions is on Hierarchical ||-coords 78 citations, the Andrienko's (GIS) 18 citations, Artero et al has 13 citations. They and almost everybody else use vertical ||-coords.

<--It is strange that the editor believes that the influence is based on the axes orientations! In fact the axes orientation is not really important, it's only related to the monitor's aspect ratio (see Moustafa & Wegman 2002,2006). More important that Wards' work is to avoid the cluttering in visualization, a problem that was first discussed in the Miller& Wegman 1991 paper. Have you ever read that article? Aslo, is the Visual computer Journal that published Inslberg's 1985 paper still exist?

So please let us agree to at least check the basic facts before making any claims. Better yet let us write an informative, useful and fun article and skip self-advertisements. I think this is right. Ainselberg 11:12, 19 March 2007 -- please write your comments separately

Contributors moved from article

I have moved this section from the article. It seems to me that if these people are important contributors then they will be mentioned as authors of references. Otherwise this section seems more appropriate in an academic paper than an encyclopedia entry. Rich257 10:37, 19 March 2007 (UTC)

Important Contributors:

---This should be corrected first and be close to the reference-- The list(chronologically) of important original work on ||-coords includes that of B.Dimsdale, J. Adams (data analysis), M. Reif, T. Chomut (data analysis), M. Boz, H. Hinterberger (density estimation 1987), J. Eickemeyer, M. Ward, A. Chatterjee, C. Gennings(biostatistics), C. K. Hung (surfaces), G. Grinstein, D. Keim, C. Jones, A. Desai, T. Matskewitch, S. & T. Avidan (data mining & classification), N. Helfman(Decision Support), A. Goel, J. Dyson(GIS), N & G. Adriendo(GIS), M. Ankerst (axes ordering), L. Yang, J. Johansson, Y. Yaari (Visualizing Complex valued functions), Y. Singer & Greenspan (Visualizing structure of large networks), A. Shaus(Visualizing derivatives).

Source for the article:

A new book has been published recently by Alfred Inselberg, see:

• http://www.amazon.com/Parallel-Coordinates-Alfred-Inselberg/dp/0817643206 , Alfred Inselberg

I don't know how to add it to the reference but maybe someone else can. Talgalili (talk) 19:53, 16 November 2009 (UTC)

A new article: Data Visualization's Final Frontier Yug (talk) 21:55, 11 August 2012 (UTC)

What is the difference between parallel coordinate plot and line chart?

Can someone explain the difference between parallel coordinates and normal line chart, like those used to display stock prices. It seems to me that what special to parallel coordinate is that it scales each column (or variable) to a common value range. —Preceding unsigned comment added by 68.147.85.229 (talk) 19:41, 8 September 2010 (UTC)

"Parallel coordinates" just means using a normal line chart in cases where the data isn't a time series (like with stock charts). The reason that there is a name and article about it is because there are some useful and non-obvious geometric properties. 64.121.106.19 (talk) 15:16, 21 May 2013 (UTC)

Additional external link to open source parallel coordinates visualization software

Hello. I read in the editor guidelines that I shouldn't add external links to sites that I maintain myself but rather let other editors decide. So here I go. I have created an open source tool that reads column-based data and displays it in parallel coordinates. The project website is www.xdat.org. I believe it provides some additional benefit to the links already provided but I guess this could be assessed more objectively by someone else. It would be great if someone would take a look and give me his thoughts on whether it's worth to be included in the list of external links. Thank you in advance! Jonny140 (talk) 22:45, 16 November 2010 (UTC)

Hello Jonny, thank you for asking instead of just introducing your link "by force". I think it is worth adding your link and will add it now. For the record - I'd wish you would look into contributing to the R project. For example, by connecting your GUI to R (and, for another example, by extending the Deducer JAVA based GUI of R, with your software). If you'd ever decide to do so - I'd be happy to mention it on my blog (http://r-statistics.com/). Best, Talgalili (talk) 10:26, 18 November 2010 (UTC)

Thank you Talgalili. Also for the suggestion to contribute to the R project. I took a long look today and I actually like the idea, although it seems like quite an undertaking for a hobby programmer like myself :-) I guess I will experiment a bit with R first, but if I decide to go further than that I will definitely let you know... Thanks again! Jonny140 (talk) 01:18, 20 November 2010 (UTC)