Most commonly shared books (weighted) WEIRDNESS
Aderi ao LibraryThing para poder publicar.
Este tópico está presentemente marcado como "adormecido"—a última mensagem tem mais de 90 dias. Pode acordar o tópico publicando uma resposta.
There are 548,832 members in LT with 36,817 user who have cataloged Harry Potter and the Sorcerer's Stone that's what 6% or 7%? So If one was incline to divide all the user of LT into randomly assigned groups of say 25 users. That would be 21,953 groups won't only about 1317 i.e. 6% of those groups have Harry Potter and the Sorcerer's Stone as the most common book? (There must be some much more torturous math there but as a rule of thumb I would guess this to be true.)
But, since the groups aren't random I would say that effect would be even less among people with special interests. For example a group like Tea! (1067 members), I bet less than 6% of them have Harry Potter and the Sorcerer's Stone.
Then look at a group like this one. I think if it was just the most common book among all our libraries it would be The elements of style which is the 79th most popular book. The second most popular non fiction book. (At least from my quick Friday afternoon scan.) the first most common non fiction book is sort of interesting.
Maybe later on the power that be would give groups more control over the statistic a give group displays. In the case of small least than average representative groups the 'Most commonly shared books (weighted):' is weird... For example the most common tag for the group would be interesting. Since this group has a particular tag 'ultb' it would be pretty interesting to see the most common tag.
Anyway these are just Friday afternoon ramblings, but I am interested in what other people think.
For one, most of those half million members have small or non-existent libraries, and those are probably the same ones who don't participate in groups. I bet way more than 6% of the members of Tea! have Harry Potter and the Sorcerer's Stone
For two, I don't follow your math at all. There's no obvious connection between the percent of users who have a book and the percent of groups which have that book as their most common. I may have to work this formally or using simulations, but I'm pretty sure it depends heavily on the distribution.
I think the key to getting 6% in each "group" is the random part. I mean if you take a random sample of just users you would get 6% with Harry Potter. I would think the size of the group would really effect the true percent.
I guess my point is really that the value of this "most common (weighted)" seem a little misleading.
I would be interesting if the owner of the groups had more control of what shows up. Or could tune this wieghting factor.
I'll be interested in what you find.
If 6% of the people have Harry Potter and the Philosopher's Stone, and every other book is ultb, then HPPS will be the most common book in every group large enough to have one. If books are randomly distributed, then every sufficiently large enough group will have HPPS as the most common book.
There is no algorithm for "most common (weighed)" that is going to get an interesting book all the time, especially with differing views of interesting. The current one produces a reasonable answer, again modulo arguments about what a reasonable answer is. And Tim Spalding, for good designing reasons, has preferred not to introduce an excess of options.
I would guess that since anyone can get an account for free there would be a lot of one time user skewing the stats. People who try it one time add a couple of books and never use it again. I wonder if there is a process of removing inactive users. I would doubt it.
Away thanks for the discussion. Maybe this thread will make what the most common book weighted is reporting clear to other user.
For that matter I don't have Harry Potter books or LotR.
The Most commonly shared books (weighted) setting is picking up the astronomy books for a couple of reasons.
1. When I started the group I looked at people who shared books with me and invited them to join the group. The user sigurd joined. He and I share these astronomy books.
2. These astronomy books are relatively rare. Bright Star Atlas has about 11 users. Nebulae and How to Observe Them (Astronomers' Observing Guides) has about 5 users and Sky & Telescope's Pocket Sky Atlas has 21 users. Of these low user numbers 2, sigurd and myself are members of this group. Since they are rarer books they are weight more. Common books for example the most common book is Harry Potter are weighted less.
3. If two people on LT are the only people on LT to have a book (you and one other) and they joined this group that book should be weighted more than the current astronomy books which are owned by 5 and up users.
4. Even if our example two people from item 3 were to join ULTB, it will take some time before this displace of most common book happened. Stats for groups do not appear to be calculated on a daily basis.
Hope that helps...
Doctor Darwin by Hesketh Pearson
You've got a friend, Charlie Brown : selected cartoons from… by Charles M. Schulz
Monitor Found in Orbit by Michael G. Coney
Isaac Asimov Presents Great Science Fiction 03 by Isaac Asimov
Star Science Fiction Stories No. 2 by Frederik Pohl
Paratime by H. Beam Piper
Carmen Miranda's ghost is haunting space station three by Sakers I really have got to find out what this is about
Knotted doughnuts and other mathematical entertainments by Martin Gardner
To keep the ship by A. Bertram Chandler
A Talent for the Invisible by Ron Goulart
These books hadn't changed since I started the group. So this is interesting...if your into odd little quirks in complete systems.
Oh, and the line is from a song - lyrics here - by Leslie Fish.