(Yet) another approach to the question of greatest directors

Introduce yourself to the community or chat with other users about whatever is on your mind
djross
Posts: 1212
5318 Ratings
Your TCI: na
Joined: Sun Apr 16, 2006 12:56 am

Re: (Yet) another approach to the question of greatest directors

Post by djross »

Deleted.
Last edited by djross on Sat Oct 15, 2011 6:52 am, edited 1 time in total.

PeaceAnarchy
Posts: 654
7005 Ratings
Your TCI: na
Joined: Tue Dec 12, 2006 1:47 am

Re: (Yet) another approach to the question of greatest directors

Post by PeaceAnarchy »

I am following Process A, I used your method, I interpreted my data and I'm attempting to revise the hypothesis based on my concerns based on my results. While the objections I presented were somewhat extreme they are based on my actual data. I really do find this method overrates someone like Spielberg, I also find it doesn't give enough credit to someone like Rohmer who has never let me down despite fewer of his films reaching my top tier. I'd express concerns with your results but I don't know enough about your tastes to say whether the list you got is reflective of your feelings on those directors, and it seems unlikely others will take the time to give more info than we have. This isn't a peer reviewed article or anything, if it were the initial statistical methods chosen would be much more robust and would have taken into account some of the objections.

I'm not sure what you expect, even if a bunch of people did this and posted lists, they wouldn't actually be useful data without an accompanying narrative by the people submitting them. It's only interesting if there's actual discussion of the results and not just waiting for more lists of names and numbers.

ShogunRua
Posts: 3449
0 Ratings
Your TCI: na
Joined: Sun May 17, 2009 3:18 am

Re: (Yet) another approach to the question of greatest directors

Post by ShogunRua »

PeaceAnarchy wrote:If the goal is to make a list of favourites, or at least a list that attempts to reflect our personal tastes then certainly a strict cutoff is not reflective of my own perceptions of directors. To take an extreme, by this system if I've watched ten masterpieces by a director and then watch his or her lesser known 20 films all of which I find terrible, that director's ranking doesn't change one bit. Yet in my own perception, my esteem for such a director would most assuredly go down, and that would happen regardless of the order in which I watched those films.


Yes, that was exactly the problem I had with the method presented in this topic.

PeaceAnarchy wrote:While ShogunRua's suggestion does alleviate this problem, it also penalizes a director who makes 10 masterpieces and 20 very good, but not great, movies, someone who I would probably give extra credit to for their large and successful body of work beyond the masterpieces. Cutting off the bottom makes both this person and the one mentioned above seem identical.


Ah, good point!

You are indeed correct; this is a special case on which my proposed alteration doesn't work effectively.

However, the original method also fails with this particular example; as you correctly note, by only taking the top 10 films, it does not matter whether the other 20 films were awful or very good, which intuitively seems wrong.

PeaceAnarchy wrote:The more I think about it the more it seems a more mathematically sound variation of what adrian mentioned might be the way to go. Perhaps some way of looking at deviations from the mean of scores and adding them up with some appropriate weighting system. Of course it's more work than I'd be willing to do by hand for an intellectual exercise, but if I had all the data in a spreadsheet I'd give some things a try to see what happens.


Yes, I agree with this. Unfortunately, the resulting formula would not be as simple or intuitive!

Shadrik
Posts: 4
1679 Ratings
Your TCI: na
Joined: Sun Aug 31, 2008 3:49 pm

Re: (Yet) another approach to the question of greatest directors

Post by Shadrik »

Interesting idea, here is my list:
1.Billy Wilder: 86,3 (7 films)
2.Alfred Hitchcock: 86,2
3.Stanley Kubrick: 85
3.Joel&Ethan Coen: 85
3.Sergio Leone: 85 (5 films)
6.Quentin Tarantino: 84,6 (7 films - excluded Sin City and Four Rooms)
7.Hayao Miyazaki: 83,8 (10 films)
8.Darren Aronofsky: 83,2 (5 films)
9.Christopher Nolan: 82,4 (7 films)
10. Woody Allen 81,3 (9 films)

Directors at 4 films who will probably enter my list soon:
Krzysztof Kieslowski: 88 (as soon as I finish The Decalogue)
Ingmar Bergman: 85
Akira Kurosawa: 83,5

Moribunny
Posts: 118
5504 Ratings
Your TCI: na
Joined: Mon Apr 03, 2006 4:48 pm

Re: (Yet) another approach to the question of greatest direc

Post by Moribunny »

Here are my top 20 following given rules, with no modifications. It's still a distorted picture but I think it's improved. I'm still not happy about giving short movies and anthology entries the same weight as features... but the result is surprising and interesting. I'm impressed by the top five! It's great to see Fassbinder, Herzog and Lumet topping this list after getting pushed down so far in my normal averages. Some oddities: Kusturica still gets lower than he deserves, Gilliam perhaps high than he deserves, and that Sirk entered the list at the expense of some of those who didn't.

1. Rainer Werner Fassbinder - 87.8 (20)
2. Werner Herzog - 87.4 (32)
3. Sidney Lumet - 86.8 (19)
4. Terry Gilliam - 86.2 (15)
5. Woody Allen - 85.4 (22)
6. Jacques Rozier - 82.7 (7)
7. Carl Theodor Dreyer - 82.2 (12)
8. Claude Chabrol - 81.7 (15)
9. Satyajit Ray - 81.3 (12)
10. Alan Clarke - 81.0 (6)
11. Emir Kusturica - 80.2 (11)
12. Ingmar Bergman - 79.4 (19)
13. Dino Risi - 79.3 (6)
14. Lynne Ramsay - 79.2 (5)
". Nicolas Roeg - 79.2 (5)
16. Mario Monicelli - 79.0 (9)
". Lina Wertmüller - 79.0 (6)
18. Luchino Visconti - 77.7 (10)
19. Ermanno Olmi - 77.6 (6)
20. Douglas Sirk - 77.1 (11)

Close: Antonioni, Fellini, Peckinpah, Costa-Gavras, Renoir, Parajanov, Kaurismaki, Scorsese

lisa-
Posts: 286
1907 Ratings
Your TCI: na
Joined: Tue Jul 30, 2013 9:22 am

Re: (Yet) another approach to the question of greatest direc

Post by lisa- »

this seems like a fun idea; i wish i had seen enough to try it.

there is some legitimacy to the concern that 10 great + 20 good = 10 great + 20 awful. for even more fun, i'd propose the following system: :lol:

Image

where n is the number of films, x is the film score, and α is a constant. the higher α is the more the average will be weighted toward the high end; α = 1 is the normal average.

it also is good for me because it doesn't require a large amount of films. using all the directors i've seen at least 4 films from:

normal average:

1. 92.33 trier
2. 91.00 bergman
3. 85.75 polanski
4. 84.22 kubrick
5. 82.00 herzog
6. 81.00 lang
7. 78.14 coen
8. 78.00 tarkovsky
9. 76.75 lynch
10. 75.25 malick

weighted average with alpha 5:

1. 93.08 trier +0.75
2. 91.42 bergman +0.42
3. 87.81 kubrick +3.59
4. 87.25 lynch +10.50
5. 86.26 polanski +0.51
6. 86.06 herzog +4.06
7. 83.08 lang +2.08
8. 82.36 coen +0.75
9. 81.20 malick +5.95
10. 79.89 tarkovsky +1.89

thanks for your time (shit, i had too much caffeine last night). :lol: :lol: i even have some code here:

http://repl.it/Lty

djross
Posts: 1212
5318 Ratings
Your TCI: na
Joined: Sun Apr 16, 2006 12:56 am

Re: (Yet) another approach to the question of greatest direc

Post by djross »

x
Last edited by djross on Thu Jul 20, 2023 9:07 pm, edited 1 time in total.

ShogunRua
Posts: 3449
0 Ratings
Your TCI: na
Joined: Sun May 17, 2009 3:18 am

Re: (Yet) another approach to the question of greatest direc

Post by ShogunRua »

lisagirl wrote:this seems like a fun idea; i wish i had seen enough to try it.

there is some legitimacy to the concern that 10 great + 20 good = 10 great + 20 awful. for even more fun, i'd propose the following system: :lol:

Image

where n is the number of films, x is the film score, and α is a constant. the higher α is the more the average will be weighted toward the high end; α = 1 is the normal average.


Not a bad try, but my problem is that the formula doesn't intuitively correspond to anything. What is alpha = 10 except an artificial method to make good movies matter way more rankings than poor ones?

Perhaps a better method is to do a weighted average of the first 10 movies with a higher coefficient alpha and then a weighted average of the rest of the movies with a lower coefficient beta.

That would also make it easier for the user to assign the level of importance they want to place on each. With your particular summation, it's not immediately clear whether someone should go for alpha = 3 or alpha = 7 based on their valuation, for instance.

lisagirl wrote:thanks for your time (shit, i had too much caffeine last night). :lol: :lol: i even have some code here:


Heh, no worries. I have done far stranger mathematical investigations on late nights before. :)

jeff_v
Posts: 81
10315 Ratings
Your TCI: na
Joined: Tue Feb 16, 2010 3:53 am

Re: (Yet) another approach to the question of greatest direc

Post by jeff_v »

I am going to try this even if it takes all day, given the number of directors involved. I'm going to smooth the curve a little though.

5-9 films seen, I'll use the overall average
10-14 seen, average of top 9
15-19, average of top 10
20-24, average of top 11
25-29, average of top 12
30-34, average of top 13
35-39, average of top 14
40+, average of top 15

ShogunRua
Posts: 3449
0 Ratings
Your TCI: na
Joined: Sun May 17, 2009 3:18 am

Re: (Yet) another approach to the question of greatest direc

Post by ShogunRua »

jeff_v wrote:I am going to try this even if it takes all day, given the number of directors involved. I'm going to smooth the curve a little though.

5-9 films seen, I'll use the overall average
10-14 seen, average of top 9
15-19, average of top 10
20-24, average of top 11
25-29, average of top 12
30-34, average of top 13
35-39, average of top 14
40+, average of top 15


Another decent try, but I can't help but think that after his ninth film, a director gets 31+ more opportunities to come up with just 6 more good films.

Post Reply