Thread Subject: Spring 2009 MATLAB Contest, April 1st-April 10th

The Spring 2009 MATLAB Contest starts today at noon EDT. It will run until April 10th. The programming challenge and full rules will be posted at the start of the contest on our website.

  http://www.mathworks.com/contest/

You can also keep up with ongoing contest announcements on our contest blog here.

  http://blogs.mathworks.com/contest/

This contest will have a very different format from previous contests. We hope you'll enjoy it and jump in and play.

Good luck!

Wow! This is a significant departure from previous contests. It'll be interesting to see how many people partake since the feedback isn't immediate and it requires us to be involved in the judging.

Is there only one final top prize? Or will you be giving prizes for something like the top 5 entries?

Do we have to publish as HTML or can we use some of the other possible formats like PPT?

Also, just to give everyone a heads up, when you download an entry on Windows, you need to completely unzip the file before viewing the published html doc in order to see the images. While Windows lets you 'browse' into compressed files like zip files without unzipping them and even lets you directly open up a html file embedded within it, it's not smart enough to also unpack the images, and thus they don't appear when you view the file. Worse yet, in Firefox you don't even get any type of icon indicating there is a broken image link to warn you of this.

At first, I didn't like this contest idea, but it has grown on me the more I've thought about it. The lack of objective scoring criteria bothered me, but the problem is actually really interesting, and I wish I had more time to work on it.

The voting system seems to me to be seriously flawed, however. For each entry, I have the option of voting for it (applying the tag) or not voting for it (not applying the tag). So, it seems to me like I should wait until all the entries are in before I vote, since there is always the possibility that I'll like the next entry better than my current favorite, and there's no way to indicate that in the voting. However, if everyone adopts this strategy, then there won't be any meaningful ranking until after all the entries are in. This would be too bad, because it would be nice to see what the community thinks of existing entries as they evolve. A simple improvement would be to have a tag for saying you like an entry (vis2009), and another tag for your final vote (vis2009_final_vote).

Looking forward to seeing what people come up with :)

-Mike Bindschadler

"Michael Bindschadler" wrote

> The voting system seems to me to be seriously flawed, however.
> For each entry, I have the option of voting for it (applying the tag)
> or not voting for it (not applying the tag).

One of the nice things about tagging is that it is perfectly reversible. You can remove a tag as easily as you add it and there is no record of it ever having existed. So I hope you will tag things that you find interesting as you go, and then you can shift around your votes as more entries come in.

Keep in mind that you can also find all the files that you personally have tagged vis2009 (as opposed to those that everyone has) by going to the advanced search page and filling out the section "Tagged vis2009 by user ID <your user ID>" And when you've done that once, you can keep it as a bookmarklet to review all of the your favorite files as needed.

See
  http://www.mathworks.com/matlabcentral/fileexchange/advanced_search

Your suggestion about vis2009_final_vote is interesting. If there is widespread support for this idea, we'd consider it... this is a very flexible contest... but I'm hoping we can just use the one tag and see how that goes.

"Ned Gulley" <gulley@mathworks.com> wrote in message <gr0lrt$oal$1@fred.mathworks.com>...
> "Michael Bindschadler" wrote
>
> > The voting system seems to me to be seriously flawed, however.
> > For each entry, I have the option of voting for it (applying the tag)
> > or not voting for it (not applying the tag).
>
> One of the nice things about tagging is that it is perfectly reversible. You can remove a tag as easily as you add it and there is no record of it ever having existed. So I hope you will tag things that you find interesting as you go, and then you can shift around your votes as more entries come in.
(snip)
> Your suggestion about vis2009_final_vote is interesting. If there is widespread support for this idea, we'd consider it... this is a very flexible contest... but I'm hoping we can just use the one tag and see how that goes.

I think it would be cool if in addition to tagging their favorites, everyone participating in this contest also rated all the submissions that they looked at, and also left a comment if they felt like it. That way a person could look at the list of submissions with the 'vis2009' tag and see what entries other people are finding well-done and interesting by sorting the search results by rating, instead of having to slog through a long list of undifferentiated entries. It would also give the authors feedback on what people liked or what they could improve on when working on their next entry, making the contest a more collaborative community effort.

a hint:

http://www.mathworks.com/matlabcentral/newsreader/view_thread/248096

us

Us,
You're forgetting that the contest was launched on April Fool's Day. It is just a matter of time until the contest team reveal that this was all a joke.
Darren.

"Darren Rowland" <darrenjremovethisrowland@hotmail.com> wrote in message <gr1dl5$c9a$1@fred.mathworks.com>...
> Us,
> You're forgetting that the contest was launched on April Fool's Day. It is just a matter of time until the contest team reveal that this was all a joke.
> Darren.

Good point, Darren. Let's wait and see. It is not the first time MWT issued such a joke if it turns out to be an one.

Yi

"Yi Cao" <y.cao@cranfield.ac.uk> wrote in message <gr1ukt$4c5$1@fred.mathworks.com>...
> "Darren Rowland" <darrenjremovethisrowland@hotmail.com> wrote in message <gr1dl5$c9a$1@fred.mathworks.com>...
> > Us,
> > You're forgetting that the contest was launched on April Fool's Day. It is just a matter of time until the contest team reveal that this was all a joke.
> > Darren.
>
> Good point, Darren. Let's wait and see. It is not the first time MWT issued such a joke if it turns out to be an one.
>
> Yi

I hope it was a joke. I'm not at all sure of that,
since Ned Gulley was one of the posters with a
submission on the FEX.

If it was a joke, I imagine they will have removed
those submissions. Even if it was a joke, we will
see MANY of these submissions appear on the
FEX now. More crap. So if it wasa joke, it was
in terribly poor taste.

John

"Ned Gulley" <gulley@mathworks.com> wrote in message <gr0lrt$oal$1@fred.mathworks.com>...
> "Michael Bindschadler" wrote
>
> > The voting system seems to me to be seriously flawed, however.
> > For each entry, I have the option of voting for it (applying the tag)
> > or not voting for it (not applying the tag).
>
> One of the nice things about tagging is that it is perfectly reversible. You can remove a tag as easily as you add it and there is no record of it ever having existed. So I hope you will tag things that you find interesting as you go, and then you can shift around your votes as more entries come in.
>

If there is a data log, then spammers, and vote-gaming can be easily detected.
If there is a behind-the-scenes scoring, it can be correlated with voting to filter fraud, and to make interesting comparisons of presentations that people like vs. presentations that are common and easy to make. If users are known and binned, their votes can be clustered based on bin-related ephemeris criteria.

It is much more than perfectly reversible. It is ingenous.

"John D'Errico" said:
> I hope it was a joke.

No, the latest contest is not a joke.

We're genuinely interested in seeing the insights that people will mine from the provided data set, and we think seeing how other people use MATLAB for the purpose of data visualization is a reasonable use of the File Exchange.

We hope it will result in the submission of some new files for the period of one week, and we look forward to learning from them. If you don't find these files interesting, you are welcome to skip past them. In any event, submissions for the contest will close on April 8th, less than a week from now.

Beyond all this, however, there is clearly a question hanging in the air: what constitutes appropriate use of the File Exchange? I'd like to move that discussion to another thread so we can continue to talk about the contest here.

For more discussion on that topic see

  http://www.mathworks.com/matlabcentral/newsreader/view_thread/248182

"Ned Gulley" <gulley@mathworks.com> wrote in message <gr2u3p$g83$1@fred.mathworks.com>...
> No, the latest contest is not a joke.
>
> We're genuinely interested in seeing the insights that people will mine from the provided data set, and we think seeing how other people use MATLAB for the purpose of data visualization is a reasonable use of the File Exchange.
>

Amidst the bombardment of negative comments and concerns about the validity of the contest, I was bit puzzled. I must admit, at first I was also bit disappointed to see the contest format. But then I realized this is because I primarily use MATLAB for numerical data processing, and expected similar sort of contests. I think MATLAB is still primarily for that. Nevertheless, I must also admit that the powerful data visualization features of MATLAB is a differentiator from other languages like C. As for myself, if I get some more time I will try to do a bit of data mining and try different visualization commands just for fun and self-training. The contest is a good motivation :)

> "Ned Gulley" <gulley@mathworks.com> wrote in message <gr2u3p$g83$1@fred.mathworks.com>...
> > No, the latest contest is not a joke.
> > We're genuinely interested in seeing the insights that people will mine from the provided data set, and we think seeing how other people use MATLAB for the purpose of data visualization is a reasonable use of the File Exchange.

Ned, aside from the inappropriate use of the File Exchange people have mentioned I find the very nature of this contest farcical. It is little more than a homework exercise,
Teacher - "Ok class, this week we'll be learning how to visualize data in Matlab. Everyone grab a copy of the Peg Solitaire contest data and create a report on the significant trends which occurred throughout the contest. Be sure to include pretty figures and justify your conclusions about the contest. Oh, and you will mark one anothers work and the best author will win a Matlab T-shirt. Also copying other people's ideas is okay."

It is a worthwhile ambition to educate people about using Matlab to display data. However I expect this kind of knowledge to be presented through the blogs and video tutorials, not via a makeshift contest. Who really has the time to devote to mining through more than 1MB of data to write a report.
You might argue that if someone has the time to submit entries to a normal contest then this is no different, but typical Matlab contests are well supported because they provide a diversion from regular work, whereas this is just like what I do day-in-day-out.
Darren

I think this contest is great.

It's neat to have access to data from a complete contest. You can figure out submission rate per person, look at a series of entries to watch algorithms evolve, differentiate between tweaks and algorithm improvements, and lots of other interesting things that I haven't thought of yet.

I'm excited to see what other metrics and trends are teased from the data. Hopefully some will be interesting and helpful enough to be included on the statistics page [1] of future contests!

Steve

[1]: Example stats: http://www.mathworks.com/contest/jumping/statistics.html

Hello,

I have a question about the following sentence in the contest rules:

"parent - This is a pointer back to a previously existing entry that this entry was cloned from. This field may be empty."

I am not sure how to interpret "pointer" here. When I examined the parent vector, I saw some entries that are above 40,000, so I guess the pointer data does not correspond to the index of a submission. What is the right way to make use of the pointer data?



"Ned Gulley" <gulley@mathworks.com> wrote in message <gr02bh$gmg$1@fred.mathworks.com>...
> The Spring 2009 MATLAB Contest starts today at noon EDT. It will run until April 10th. The programming challenge and full rules will be posted at the start of the contest on our website.
>
> http://www.mathworks.com/contest/
>
> You can also keep up with ongoing contest announcements on our contest blog here.
>
> http://blogs.mathworks.com/contest/
>
> This contest will have a very different format from previous contests. We hope you'll enjoy it and jump in and play.
>
> Good luck!

Hi,

Can I have some clarification about what it means to "clone" someone else's code? Namely, what does it mean when the "parent" variable is nonzero?

I have noticed pairs of Peg Solitaire entries which have extremely similar program code, and yet the program submitted later in the pair claims to have no parent.

"Ned Gulley" <gulley@mathworks.com> wrote in message <gr02bh$gmg$1@fred.mathworks.com>...
> The Spring 2009 MATLAB Contest starts today at noon EDT. It will run until April 10th. The programming challenge and full rules will be posted at the start of the contest on our website.
>
> http://www.mathworks.com/contest/
>
> You can also keep up with ongoing contest announcements on our contest blog here.
>
> http://blogs.mathworks.com/contest/
>
> This contest will have a very different format from previous contests. We hope you'll enjoy it and jump in and play.
>
> Good luck!

Hi,

Can I have some clarification about what it means to "clone" someone else's code? Namely, what does it mean when the "parent" variable is nonzero?

I have noticed pairs of Peg Solitaire entries which have extremely similar program code, and yet the program submitted later in the pair claims to have no parent.

"Ned Gulley" <gulley@mathworks.com> wrote in message <gr02bh$gmg$1@fred.mathworks.com>...
> The Spring 2009 MATLAB Contest starts today at noon EDT. It will run until April 10th. The programming challenge and full rules will be posted at the start of the contest on our website.
>
> http://www.mathworks.com/contest/
>
> You can also keep up with ongoing contest announcements on our contest blog here.
>
> http://blogs.mathworks.com/contest/
>
> This contest will have a very different format from previous contests. We hope you'll enjoy it and jump in and play.
>
> Good luck!

Two questions / concerns I have:

1. I just submitted an entry, and was surprised to NOT see it immediately appear online. Instead I got a message saying it needed to be reviewed and would take up to 5 days to appear?!?!?! I don't even have any type of link or reference to the upload listed in my profile, so for all I know it will disappear into the ether. Who is doing the reviewing? Obviously this could significantly impact the ability of entries to get seen by others if they aren't made public in a short amount of time.

2. My entry happens to include a video (shameless plug.. go see it please.. it's cool;). I figured out a way to make MATLAB create an AVI file and then display it as part of the published HTML. For this first submission, I deliberately kept the animation small and short, but the zip file is still over 2.5MB and uncompresses to over 150MB. Do we really need to include 'all the published' files? Downloading the .m file and running / publishing it produces the same results without taking up so much bandwidth.

"Alan Chalker" <alancNOSPAM@osc.edu> wrote in message <grc66b$dd0$1@fred.mathworks.com>...
> Two questions / concerns I have:
>
> 1. I just submitted an entry, and was surprised to NOT see it immediately appear online. Instead I got a message saying it needed to be reviewed and would take up to 5 days to appear?!?!?! I don't even have any type of link or reference to the upload listed in my profile, so for all I know it will disappear into the ether. Who is doing the reviewing? Obviously this could significantly impact the ability of entries to get seen by others if they aren't made public in a short amount of time.

> 2. My entry happens to include a video (shameless plug.. go see it please.. it's cool;). I figured out a way to make MATLAB create an AVI file and then display it as part of the published HTML. For this first submission, I deliberately kept the animation small and short, but the zip file is still over 2.5MB and uncompresses to over 150MB. Do we really need to include 'all the published' files? Downloading the .m file and running / publishing it produces the same results without taking up so much bandwidth.

These two problems strongly support the notion that it was a poor idea in the first place to use the File Exchange for this contest ...

Jos

"Alan Chalker" <alancNOSPAM@osc.edu> wrote in message <grc66b$dd0$1@fred.mathworks.com>...
> Two questions / concerns I have:
>
> 1. I just submitted an entry, and was surprised to NOT see it immediately appear online. Instead I got a message saying it needed to be reviewed and would take up to 5 days to appear?!?!?! I don't even have any type of link or reference to the upload listed in my profile, so for all I know it will disappear into the ether. Who is doing the reviewing? Obviously this could significantly impact the ability of entries to get seen by others if they aren't made public in a short amount of time.
>
> 2. My entry happens to include a video (shameless plug.. go see it please.. it's cool;). I figured out a way to make MATLAB create an AVI file and then display it as part of the published HTML. For this first submission, I deliberately kept the animation small and short, but the zip file is still over 2.5MB and uncompresses to over 150MB. Do we really need to include 'all the published' files? Downloading the .m file and running / publishing it produces the same results without taking up so much bandwidth.

A quick update: The file was 'approved' at 8:30AM today. If you are interested in seeing it: http://www.mathworks.com/matlabcentral/fileexchange/23565 (I know, another shameless plug, but I figure this contest is also about marketing your submissions;)

Maybe I'm missing something, but this contest seems difficult to keep tabs on if you are not actively involved. Is there a way to see the published M-files without downloading the data set and files and then running them? There might be greater voter participation if you could.

"Nicholas Howe" <NikHow@hotmail.com> wrote in message <gri8hh$8r8$1@fred.mathworks.com>...
> Maybe I'm missing something, but this contest seems difficult to keep tabs on if you are not actively involved. Is there a way to see the published M-files without downloading the data set and files and then running them? There might be greater voter participation if you could.

Nick,

If you go to the file information about each file, there is a link to the published m-file. You can see the entry there without any downloading.

Doug

Do not have time for real submission. Just played with data for fun.
Some interesting results/figures.
1. Create binary matrix “line in entry” (see Nathan code)
2. Run primary component analysis
3. Run clustering analysis

See you in fall.

SY

> Nick,
>
> If you go to the file information about each file, there is a link to the published m-file. You can see the entry there without any downloading.
>
> Doug

Oops! Too bad I didn't know this. I submitted an entry with new code, but I unwittingly left in place the old published m-file from the entry I had based mine on. So when people look at it, they won't see any difference from the previous entry. Sorry about that!

For what it is worth, I updated the published m-file in my submission (after the noon deadline). At least this way people will get to see it as it was intended.

I also submitted a late entry which I think has some interesting results concerning tweaks vs. innovations.

Both changes should be visible once they are approved.