Display rating hisotogram

[mike_morgan1]

<p>As of the time of writing, <a

href="http://www.photo.net/photo/2168910">this

photo</a> has received the ratings indicated in the histogram

below.</p>

<p>Automatically generating and displaying this type of histogram

automatically

for a photo in a user's folder might not be too difficult, and might

put

excessively high or low scores in perspective.</p>

<p>Anyway, I think it is an interesting way to present the ratings a

photo has

garnered.</p>

<p> </p>

[sammm]

I think this is very interesting and useful, but does contributing that much space on a page to a histogram rather than comment or discussion put too much emphasis on the ratings?

 

My suspicion is that most photos will show a curve around the average rating, and that the most interesting histograms would be ones that show two lumps (e.g., some get it, some don't) or the few that show the "piling on" of 6s and 7s (with attendant 2 word effusive but useless comments).

[james_oneill]

Mike, why did you name your son Nikon ?

 

OK, it was a stop, but my eye saw a comma.

 

Actually it is very interesting, and I do that with my portfolio.

 

This is a graph of the average scores e.g. 3 pics have averages of 4.5, 9 have an average of 4.6 etc.

 

I'd love to be able download the detail, not just a summary as a CSV file.<div></div>

[maxz]

<P>Sam,</P><P>Today we have tabs on the page with the photo (the gray area). Why not have a rating histogram tab? Then it wouldn't take any extra space, and you would have to click an extra time just to see it.</P><P>You could extend this as James did to show the distribution of a folder or entire portfolio of a member. Or the other way around as I proposed in an <A HREF="http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=007VLK">earlier thread</A> (look at the end of the page), to show how a member rates other photos. This might be interesting to the member him-/herself to see if they tend to deviate from the "standard"/expected rating distribution or for other members wanting to know if a person is a notorious over-/underrater.</P><P>Graphically displaying data is often very informative and usually tells us a lot more than just a bunch of raw numbers (as the ratings are displayed today).</P><P>Any graphs should be non-intrusive and IMHO only available on request (ie not pop up in your face). After all, this is Photo.Net and not Diagram.Net</P>

[gordonr]

It's called "chartjunk" - graphical information with limited content or meaning. See: The Visual Display of Quantitative Information by Edward R. Tufte, or his website http://www.edwardtufte.com/

[maxz]

<BLOCKQUOTE><I>It's called "chartjunk" - graphical information with limited content or meaning.</I></BLOCKQUOTE>

<P>Limited content or meaning? Of course it is, this is not numbers we would base national economy on, it is just a bunch of ratings on Photo.Net and without a context any chart is junk. But it is clearly more informative than the <I>Viewed 1002 times with 6 ratings Aesthetics 4/7 Originality 4/7</I> that we have today. Does the 4/4 rating mean that we have 3 1/1:s and 3 7/7:s which would indicate bogus ratings or something else? Or worse the <A HREF="http://www.photo.net/photodb/ratings_breakdown?photo_id=2168910">Ratings breakdown page</A> whose only use today seem be to find the high/low raters and reward/retaliate.</P><P>For me it would not be "chartjunk" since I could be able to read out more information from this graph quicker than from the ratings breakdown list. So why not give Tufte some extra material for his next book?</P>

[mike_morgan1]

<p><p>Today we have tabs on the page with the photo (the gray area). Why not have a rating histogram tab?</i><p><p>

 

Yes, good idea.<p><p>

<i>It's called "chartjunk" - graphical information with limited content or meaning. </i><p><p>

If you want to argue that ratings in of themselves have limited meaning and content, I won't attempt to argue.<p><p>

But since they are there, displaying the distribution of ratings is actually more useful than a tabular display, in my opinion.

[john falkenstine]

Histograms and charts are often used in corporate meetings for "dog and pony shows" and to snow incompetent corporate management. (Dilbert cartoons, unfortunately, are very true) When I see folks display them proudly, I usually mentally flag that party as somebody to avoid and to give them as little information as possible. Quality control freaks also love to use them. I have never found them to be of any use unless the conditons under which they are used are accurately defined. In the case of the photo.net ratings they look cool but are essentially meaningless.

[mike_morgan1]

I don't think they look cool, and I think they are meaningful. And anybody who disagrees with me should be the subject of a Dilbert cartoon.

 

Now that we've heard from some in the ad hominem crowd, anybody got actual reasons.

[john falkenstine]

ad hominem: Not to worry, you put forth a question and you will get pro/con replies. I gave you my impression of these charts. I don't believe in personal attacks-photo.net is for the enjoyment of photography. Regards.

[davidweaverphoto]

Chartjunk...HAHAHAHAHHAHAHA. Yeah, I'm a big fan of Tufte's but a simple histogram is a simple and elegant method to display this info. What woud you suggest? Box charts? Happy faces that somehow represented a distribution? <BR>

Maybe substitute various states of the penis for the "nudes" ratings? Hahaha...go back and reread all of Tufte's books, you obviously missed his spendid and elegant points. The original poster states a reasonable question. Thanks for the good laugh, BTW.

<BR>

LOL - David

[mike_morgan1]

I suppose if all of the charts ended up looking like this, they wouldn't be too useful. But it's a thought.

[gordonr]

The simplest response is the old programmers saying: "garbage-in, garbage-out".

 

Statistics 101 should teach that the closer you look at "random" data, the less meaningful it becomes...

[james_oneill]

Without naming the (very well known) organisation I work for I have to sit through various torture by powerpoint sessions, and read long "communications" where the poorly articulated message is reinforced or contradicted by a chart. Having studied statistics in the later stages of school and as part of my university degree I have a tendency to pull people's numbers appart. My dear employer accuses me of having a negative attitude because I pointed out that when they say they want me to spend X days on this, Y days on that, and Z days on the other, and still put W% of my time into my main task, that would only be possible in a year of about 420 days.

Somewhere I've got Huff's "How to lie with statistics". <p>

So all in all I don't think I need to be told about bad charts and stats. (And <b>Mike</b> the pedant in me needs to say these are column graphs, histograms are charts which use lines of icons. Whoever named those charts of brightness histograms did the language a dis-service. But the people can't use "presently" to mean "at present" so ....)<p>

 

My point is I do analyse my stats quite a lot. I've got 1667 ratings for my current pictures, and about another 100 for deleted ones. I keep them in a spread sheet and have an access database which has queries like <br>

<i>The fan club</i>: People who have given the top scores to two or more pictures.<br>

<i>The Knockers</i>: People who have given the bottom scores to two or more pictures.<br>

<I>Whose top rated pictures</i>: People with more than 2 pictures scored 6/6 or better.<br>

<i>Weighted scores</i> Basically, average * log(count of ratings)<br>

<i>No Min max</i>: Averages with the highest and lowest excluded<br>

The last two get rolled into my main picture scores query, together with information about proportion of 6s and 7s and pictures with no scores below 5.<p>

I use three charts. <br>

Distribution of Originality scores<br>

Distribution of Aesthetics scores<br>

Distribution of picture averages (above)<br>

I have enough data to be statistically relavent, and the charts do help me to understand the data. <p>

 

I wouldn't suggest the site provides all these things directly. But allowing the download of a CSV file with ALL my scores in would be great (the present summary allows me to find out which pictures have picked up scores an go and screen scrape the data). I'd be happy to share my spreadsheet and database for others to use as a template.

[mg]

Was it Montaigne who wrote "Science without a conscience is only ruin of the soul"...? :-)

<p>

This thread is quite adorable, or cute to my eyes. It shows me how much expectations photographers put in the proper understanding of the "feedback" they get through numbers, whereas the Editor of the site has repeated number of times that the ratings are not feedback to the photographers - the comments are.

<p>

This thread certainly shows that those who posted here do interprete the ratings as feedback, despite Bian Mottershead's explanations. That's an interesting observation imo, because it leads me to question who is right here and who is wrong. Are ratings feedback to the photographer, yes or no? I have an opinion, but I'll keep it to myself. :-)

<p>

Now... MORE interesting... And please pardon me the sarcasm, I don't mean to be rude at all, but I just can't resist humor sometimes...:-)

<p>

May I just ask: what value would you give to ordered numbers, O Masters of Statistics, if the figures you use in your graphs and charts, pies, histograms and cheese-cakes of all sorts, IF these figures would be the result of an addition of salad leafs, potatoes, quantities of water, and number of salt and sugar grains...? :-))

<p>

If you read the sacasm here, what I mean is this... Each number in the little columns means something different. The scales used to rate by various raters are almost never the same. So how can you compare them, and moreover, add them ?!? On top of that, have you began to realize that a large percentage of the high ratings we all got are from people who expected some favor from us...? Have you wondered also who really looked at your picture carefully enough to actually see potential technical or artistic misgivings...? If not, I suggest you start thinking about it. Perhaps after that, you'll get enough creative inspiration to invent the flower-shape histogramm that will reveal peace instead of war...:-))

<p>

Nevertheless, I have to say that artistically speaking, your graphs look quite good. Like a picture of a treeless hill generated in PS... Unfortunately, in terms of content and originality, I'm affraid my rating would be low, very low...:-)

[n_p]

Please don't lump me in with the ad hominem crowd, but I too find it fascinating how much value many attribute to the stupid numbers game that takes place on this site. If you're a statistician, maybe this is an interesting excercise indeed, but really, the quality photographic discourse on this site has been dead for a long time now.<br>My portfolio is full of photo.net metaphors. My problem entirely, of course.