Comments for Justin Talbot http://www.justintalbot.com Tue, 24 Aug 2010 18:59:59 -0400 hourly 1 http://wordpress.org/?v=3.0 Comment on Axis Labeling by An Extension of Wilkinson’s Algorithm for Positioning Tick Labels on Axes – Justin Talbot http://www.justintalbot.com/research/axis-labeling/comment-page-1/#comment-135 An Extension of Wilkinson’s Algorithm for Positioning Tick Labels on Axes – Justin Talbot Tue, 24 Aug 2010 18:59:59 +0000 http://www.justintalbot.com/?page_id=154#comment-135 [...] Axis Labeling [...] [...] Axis Labeling [...]

]]> Comment on An Extension of Wilkinson’s Algorithm for Positioning Tick Labels on Axes by Justin Talbot http://www.justintalbot.com/2010/07/an-extension-of-wilkinson%e2%80%99s-algorithm-for-positioning-tick-labels-on-axes/comment-page-1/#comment-134 Justin Talbot Tue, 24 Aug 2010 18:57:19 +0000 http://www.justintalbot.com/?p=157#comment-134 You're right that more white space would look better. This was an issue that I ran out of time and space to address correctly. Basically, it's not sufficient to just add 4% whitespace to the resulting axis because the optimization function might have found a better solution had it known that it had more space to work in. On the other hand, you can't just pad the white space up front, because you want the ending ticks to be close to the actual data extremes, not to the padded extremes. To address this, our coverage component should probably be split into two parts, one evaluating white space and the other evaluating closeness to the extrema. Good future work. I'm not exactly sure what you mean by the space between the first/last tick and the axis. Do you mean the space between the first and last tick and the respective end of the axis *line*? If so, that would be a good change (another adjustment to the coverage criteria) and should be easy for plots with lots of labels. However, for plots with few labels, my guess is that this would introduce a lot of extra white space. You’re right that more white space would look better. This was an issue that I ran out of time and space to address correctly. Basically, it’s not sufficient to just add 4% whitespace to the resulting axis because the optimization function might have found a better solution had it known that it had more space to work in. On the other hand, you can’t just pad the white space up front, because you want the ending ticks to be close to the actual data extremes, not to the padded extremes. To address this, our coverage component should probably be split into two parts, one evaluating white space and the other evaluating closeness to the extrema. Good future work.

I’m not exactly sure what you mean by the space between the first/last tick and the axis. Do you mean the space between the first and last tick and the respective end of the axis *line*? If so, that would be a good change (another adjustment to the coverage criteria) and should be easy for plots with lots of labels. However, for plots with few labels, my guess is that this would introduce a lot of extra white space.

]]> Comment on Axis Labeling by Ahmet Engin Karahan http://www.justintalbot.com/research/axis-labeling/comment-page-1/#comment-133 Ahmet Engin Karahan Sun, 22 Aug 2010 16:32:34 +0000 http://www.justintalbot.com/?page_id=154#comment-133 Thank you for your relatively rush answer. It clears up my questions some degree. I use a helper function v() in my implementation and that's why I never thought that these parameters required to calculate v(). My implementation determines v as following: public int v() { return (offset == 0 || offset-stepSize == 0) ? 1 : 0; } I have "i" already available, and do not require q. In my current work I've done with the nice step size generator, and tested it comparing with Table 1 in a unit test. And I had trouble with search algorithm as I wrote here. If you' evaluate the search algorithm within your code, please send me this section earlier. I'll send you my Java implementation when I've done with it, so the people benefit both of them approximately at same time. Thank you again. Thank you for your relatively rush answer. It clears up my questions some degree. I use a helper function v() in my implementation and that’s why I never thought that these parameters required to calculate v(). My implementation determines v as following:
public int v() {
return (offset == 0 || offset-stepSize == 0) ? 1 : 0;
}

I have “i” already available, and do not require q.
In my current work I’ve done with the nice step size generator, and tested it comparing with Table 1 in a unit test. And I had trouble with search algorithm as I wrote here. If you’ evaluate the search algorithm within your code, please send me this section earlier.
I’ll send you my Java implementation when I’ve done with it, so the people benefit both of them approximately at same time.
Thank you again.

]]> Comment on Axis Labeling by Justin Talbot http://www.justintalbot.com/research/axis-labeling/comment-page-1/#comment-132 Justin Talbot Sun, 22 Aug 2010 15:33:11 +0000 http://www.justintalbot.com/?page_id=154#comment-132 Sorry you're having trouble; you point out some valid problems with the description in the paper. I'll try to get my code out quickly since that would be the clearest answer. However, I'll try to address your specific questions. The simplicity, density, and coverage functions in the inner loop are the functions defined in sections 4.1-4.3. For your question about the parameters, I agree that the paper is not completely clear. However, the parameters are used. For example, in the simplicity function I use lmin, lmax, and lstep to determine the value of *v* (whether or not the sequence includes 0). The variable q is the element of Q being used to generate the sequence and it is necessary to compute *i* (e.g. q = Q[i]). However, depending on how you iterate through Q in the outer loop, you may have i already available and you can pass it in directly. The simplicity_max, density_max, and coverage_max functions used in the outer loops are approximations to the actual functions. They are used to restrict the search space, but not in the final score (that's why the actual functions have to get called in the inner loop). The approximations are described in the text of section 5. This, I think, is one of the harder things to get directly from the paper. i does start at 1. I doubt that starting at 0 would have a large impact on the results, but that is a detail I should have included. Good luck on the implementation. I'll let you know when my code is out. Sorry you’re having trouble; you point out some valid problems with the description in the paper. I’ll try to get my code out quickly since that would be the clearest answer. However, I’ll try to address your specific questions.

The simplicity, density, and coverage functions in the inner loop are the functions defined in sections 4.1-4.3. For your question about the parameters, I agree that the paper is not completely clear. However, the parameters are used. For example, in the simplicity function I use lmin, lmax, and lstep to determine the value of *v* (whether or not the sequence includes 0). The variable q is the element of Q being used to generate the sequence and it is necessary to compute *i* (e.g. q = Q[i]). However, depending on how you iterate through Q in the outer loop, you may have i already available and you can pass it in directly.

The simplicity_max, density_max, and coverage_max functions used in the outer loops are approximations to the actual functions. They are used to restrict the search space, but not in the final score (that’s why the actual functions have to get called in the inner loop). The approximations are described in the text of section 5. This, I think, is one of the harder things to get directly from the paper.

i does start at 1. I doubt that starting at 0 would have a large impact on the results, but that is a detail I should have included.

Good luck on the implementation. I’ll let you know when my code is out.

]]> Comment on Axis Labeling by Ahmet Engin Karahan http://www.justintalbot.com/research/axis-labeling/comment-page-1/#comment-131 Ahmet Engin Karahan Sun, 22 Aug 2010 11:51:11 +0000 http://www.justintalbot.com/?page_id=154#comment-131 Hi, I'm working on a Java implementation of your paper "extended Wilkinson axes labeling". However in your pseudo code of searching algorithm I could not decipher the different notations of the same functions. You gave a definition of simplicity, density and coverage, and I wrote functions with appropriate parameters. But what about the functions of the same name in the inner most loop of your pseudo code. What you mean for example with a simplicity calculation simplicity(q,j,lmin,lmax,lstep) I can calculate it with Q, i, j, v() as you defined in the paper, What about the unneccary parameters q (if it's not Q), lmin, lmax, lstep. Same questions about the other two... Please clarify all such things to be able to usefull. I think your article of Extented Wilkinson algorithm is full with excellent ideas, but it hides a lot due to strange notations and extremely terse sentences. One more example the parameter i. The first question was if it starts from 0 (like C/Java) or from 1. The answer is not found in your paper, but in Wilkinson's book of GoG. Hi,
I’m working on a Java implementation of your paper “extended Wilkinson axes labeling”. However in your pseudo code of searching algorithm I could not decipher the different notations of the same functions. You gave a definition of simplicity, density and coverage, and I wrote functions with appropriate parameters. But what about the functions of the same name in the inner most loop of your pseudo code.
What you mean for example with a simplicity calculation simplicity(q,j,lmin,lmax,lstep) I can calculate it with Q, i, j, v() as you defined in the paper, What about the unneccary parameters q (if it’s not Q), lmin, lmax, lstep. Same questions about the other two…
Please clarify all such things to be able to usefull. I think your article of Extented Wilkinson algorithm is full with excellent ideas, but it hides a lot due to strange notations and extremely terse sentences. One more example the parameter i. The first question was if it starts from 0 (like C/Java) or from 1. The answer is not found in your paper, but in Wilkinson’s book of GoG.

]]> Comment on Axis Labeling by Justin Talbot http://www.justintalbot.com/research/axis-labeling/comment-page-1/#comment-130 Justin Talbot Thu, 19 Aug 2010 01:05:36 +0000 http://www.justintalbot.com/?page_id=154#comment-130 I don't have a Java implementation at the moment. I'm preparing a C# implementation for release in the next few weeks. I have a limited R version working as well. It should be pretty straightforward to port the C# implementation to Java. I'll let you know when it's available. I don’t have a Java implementation at the moment. I’m preparing a C# implementation for release in the next few weeks. I have a limited R version working as well. It should be pretty straightforward to port the C# implementation to Java. I’ll let you know when it’s available.

]]> Comment on Axis Labeling by Thomas Freeman http://www.justintalbot.com/research/axis-labeling/comment-page-1/#comment-129 Thomas Freeman Wed, 18 Aug 2010 15:17:50 +0000 http://www.justintalbot.com/?page_id=154#comment-129 Hi Justin, I just read your paper on axis labeling and was very impressed by the results of the algorithm. I am a recent Biochemistry Ph.D. and am interested in developing a specific type of scatter plot program. I have no background in CS other than what I've been able to teach myself. I am writing to ask if your axis labeling code will be available in an open source Java implementation? I would like to try to implement your code in my program if possible. Thanks, and good luck with getting your Ph.D.! Hi Justin,
I just read your paper on axis labeling and was very impressed by the results of the algorithm. I am a recent Biochemistry Ph.D. and am interested in developing a specific type of scatter plot program. I have no background in CS other than what I’ve been able to teach myself. I am writing to ask if your axis labeling code will be available in an open source Java implementation? I would like to try to implement your code in my program if possible. Thanks, and good luck with getting your Ph.D.!

]]> Comment on An Extension of Wilkinson’s Algorithm for Positioning Tick Labels on Axes by Hadley Wickham http://www.justintalbot.com/2010/07/an-extension-of-wilkinson%e2%80%99s-algorithm-for-positioning-tick-labels-on-axes/comment-page-1/#comment-123 Hadley Wickham Sat, 31 Jul 2010 14:02:45 +0000 http://www.justintalbot.com/?p=157#comment-123 The R graphics don't look quite right - they're missing the standard 4% offset between the data and the axes. I also wonder if you could make the ticks more aesthetically pleasing by making the space between the first/last tick and the axis more symmetric. The R graphics don’t look quite right – they’re missing the standard 4% offset between the data and the axes.

I also wonder if you could make the ticks more aesthetically pleasing by making the space between the first/last tick and the axis more symmetric.

]]> Comment on our day, visualizing the American Time Use Survey by Rich Tatum http://www.justintalbot.com/2009/02/our-day-visualizing-the-american-time-use-survey/comment-page-1/#comment-100 Rich Tatum Thu, 17 Dec 2009 04:25:02 +0000 http://justintalbot.org/?p=85#comment-100 Justin, what tool did you use to create the visualilzation? Tableau? Adobe Flare? Regards, Rich Justin, what tool did you use to create the visualilzation? Tableau? Adobe Flare?

Regards,

Rich

]]> Comment on Gallup-Healthways Well-Being by Geithner gets tough questions from Senate panel (Reuters) — But As For Me http://www.justintalbot.com/2009/03/gallup-healthways-well-being/comment-page-1/#comment-21 Geithner gets tough questions from Senate panel (Reuters) — But As For Me Fri, 13 Mar 2009 02:07:51 +0000 http://justintalbot.org/?p=89#comment-21 [...] Gallup-Healthways Well-Being - Justin Talbot [...] [...] Gallup-Healthways Well-Being – Justin Talbot [...]

]]>