Track popularity by date

Tim whipped up a fun feature overnight: a way to track a work's popularity over time.

Check it out at:

http://www.librarything.com/zeitgeist/popularity

More on the blog.

This is fantastic! To some degree you can tell what's coming out in the theaters or on TV by what's rising fast. But then others like Bridge to Terabithia are more puzzling.

Yeah Tim and I spent a lot of time today trying to figure out what the heck was going on with some of them!

Movies are definitely an interesting trend to watch - school starting also seems to have an effect on some titles.

For those works in the same position as the previous month, maybe a "No movement" icon would be something to consider.

Ah, that'd probably account for Terabithia and some of the others.

Oh, and I meant to mention - the book images really make this feature. It would be interesting data, but to me being able to see the books make it more "fun".

Neat! I already found one book I'm adding to my wishlist (A Discovery of Witches) while I was comparing 2010 and 2011. It made a gigantic leap and so I had to see what it was about, so I checked it out and it sounded interesting.

Nice.

Is there any chance of getting a similar thing that shows popularity of books published in a given year? Basically this Goodreads feature.

>7 _Zoe_: Hmm. Doable, I think, but not today :-)

Oh well, maybe tomorrow ;)

OK, I think I understand that this is the rank of the number of new books of each work added to member libraries in the time period. I think this would be more understandable if you included the raw number in there somewhere and spell out what the number is Something like "200 books added in August".

For the August 2011 list why is the arrow next to Harry Potter and the Prisoner of Azkaban red and the one next to Harry Potter and the Order of the Phoenix yellow since their both -2?

#11 by jjwilson61> I'm wondering if it's done as a percentage of popularity. PoA was the #2 most popular work, OotP was #5. So #3 dropping two points is a higher percentage than #5 dropping two points.

Of course, looking at some of the others I don't think that fully explains it, but I'm betting it's headed down the right path.

>11 jjwilson61: - If I remember what Tim told me correctly when I asked this question, the raw number decrease (by percentage or something) is greater for the red one than the yellow one, even though they both fell the same number of ranks. I don't know that we'll add the raw numbers into the display (we had them in for admins at one point, but decided it didn't look very nice).

What's up with this one?

http://www.librarything.com/work/9293902/popularity

Note all the blank boxes (in between the boxes with numbers) and according to the CK A Dance With Dragons wasn't even released until this year.

Hm - that's definitely odd. And yes, DwD is the brand-new one. Maybe something's gotten combined in with it?

Yeah, something weird's def. going on there. Not immediately evident at first glance, but very odd.

Hasn't the title for DwD been known for years and years? almost since the last book was written? I can imagine that die-hard fans have been adding it to wishlists for that long (though I dunno how they'd be doing it without a source...) (since not everyone would bother with the manual add)

ah, that may account for it too, actually!

Has it been known since September 2005? Because that's how far back the numbers in the last chart go.

Well, the previous book, which did note the title of the next volume, was published sometime in the fall of '05, if memory serves ... so it's theoretically possible ...

I see that volumes 6 and 7 have already been entered in LT; see the series page.

>14 jjwilson61:-16

Actually, that doesn't look all that wrong to me - A Dance with Dragons was "supposed" to come out back in 2005/2006 so I wouldn't be surprised if a lot of people had it on wishlists or preordered or in their catalogues in some form or another since then. The "blanks" are just a result of it falling below the threshhold. The definite release date was given earlier this year (looking at that page, probably March) and a lot of people would have ordered it/added it to a wishlist/tbr list then as well.

>20 jbd1:

Yeah, and the afterword suggested that Dance with Dragons would be out about 6 months later.

I think the DWD problem is solved. Wishlists and hope springing eternal...

Publication date isn't currently tracked at the work level in the same way. I want to do it at some point.

From my perspective the missing piece is the "hot this month" data.

I demand "hot this minute"!

What do all the different arrows mean? I know that generally an up-arrow means the popularity has gone up, and a down-arrow means the popularity has gone down....

But I'm seeing green and yellow up-arrows, as well as yellow and red down-arrows. So.... yellow is both up and down?

Green: up by a lot
Yellow: up or down by a little
Red: down by a lot

I think...

Publication date isn't currently tracked at the work level in the same way. I want to do it at some point.

What do you mean? Is the CK data not useable?

>26 justjim:

Right.

>27 _Zoe_:

It's usable, but covers only a tiny fraction of titles (1-2%). We need to grab that info from the Amazon records and lit it to the work level.

>28 timspalding: Ugh. Won't that ruin it by having the dates of new editions instead of when the book first came out?

People will enter more data if the field does something, and I suspect that the most popular are also the 1-2% that have data already.

>28 timspalding:
Why not take the earliest book level publication date associated with a work as the work original publication date and then let people override with the canonical date to correct errors.

>28 timspalding:

No, you take the first date. Library records also have the data. You need some common-sense logic in the case of classics—probably don't make a guess at all if the spread of dates is more than 50 years. But 99% will be trivial.

>29 _Zoe_:

Right.

>31 timspalding: But don't you think there's a massive, massive overlap between the most popular books and the ones that have original publication dates entered?

It just seems like a waste to have a field for the precise data you want but not even try to use it.

>32 _Zoe_:

More generally, I want LT to say "Such and such a book by such-and-such an author (1973)." Sure, it's nice for popular books, but I think it's nice generally.

>33 timspalding: I agree that that would be nice. I also think if you added that info and drew it from original publication date, you'd get thousands of dates added overnight. You always underestimate your users.

Heh. New time sink. ;-)

So many sinks, so few fountains...

I've noticed a few books without numbers or arrows and I've assumed that means that there was no previous change from the prior year. However, I think that might be confusing if, for example let's compare 2010 and 2011 books, “The Hobbit” at #16 has no arrow or number in 2010 category but is at #13 in 2011. Only if we had compared the 2009 and 2010 books would we have noticed that the book stayed at the same place. Perhaps there should be a message saying the book stayed at the same place? Also, for "The Hobbit" in 2011, the arrow says it went up two place from 2010 when in fact it has gone up three.

#37 by macart3> I don't think it's that confusing when showing only 2010 and 2011 to figure out that the 2010 arrows are based off of a change from 2009.

But in any case, I'd favor side-to-side arrow icon, similar to this:

Small changes don't produce an arrow. Bigger ones do. Really big ones produce a green or red.

I'd like to be able to adjust both the left and right column so I could decide what months/quarters/years I was comparing. I suppose that could create a lot of on the fly calculations that might bog down the servers, but it would be interesting to see things like the popularity difference between The Deathly Hallows at its publication date (2007) vs. when the first Deathly Hallows movie was released in 2010.

New feature:
http://www.librarything.com/topic/122881