Surprise Marriage
- Thread starter MangaDex
- Start date
Dex-chan lover
- Joined
- Dec 31, 2019
- Messages
- 119
OKAY!!! DEFINITELY WAS NOT EXPECTING, NOR WAS I READY FOR HER TO CALL HIM THE N-WORD, BUT JUST ROLL WITH IT, I GUESS??????
- Joined
- Nov 3, 2019
- Messages
- 320
ah classic chinese manhua, what is wrong with these kind of characters? lol always drama, forced kiss etc jeez that women better learn karate, a strong female lead is always good against a rich piece of shit!
Dex-chan lover
- Joined
- Dec 31, 2019
- Messages
- 119
Was this picked up and reuploaded by another group? If so, is it a different translation than the last one, because in the last one,
and let me say, I was NOT ready for that.
Well, she doesn't say it outright. She calls him that in her head, but still.
she literally calls her fiance the n-word
Well, she doesn't say it outright. She calls him that in her head, but still.
- Joined
- Jun 9, 2018
- Messages
- 2
No, this isn't Shoujo ai. It might be Shoujo, but no.
- Joined
- Nov 29, 2018
- Messages
- 109
Dropped. What kind of idiotidiotic mess is this?
- Joined
- Mar 18, 2019
- Messages
- 113
Why are therentwo dofferent rating on each manga and which is the more reliable one? They aren't usually this different
Dex-chan lover
- Joined
- May 20, 2018
- Messages
- 3,426
I am guessing the rating difference is because people on that site are used to an expecting Chinese drama.
Dex-chan lover
- Joined
- Nov 20, 2018
- Messages
- 5,157
@kingofhell99, @starburst98—
The first number is an attempt at Bayesian averaging; the second is the ordinary average of individual scores.
A Bayesian rating starts with what is called a “prior mean” or just “prior”, and a weighting; it is as if a bunch of folk have already scored the series. When the first actual scores are entered, they begin to count in the calculation; over time, the relative significance of the prior and its weighting become ever smaller.
If the prior is reasonable, and if the weighting is appropriate, then a few yahoos don't cause a series to be scored unreasonably high or low. If the weighting is too low, then the prior has almost no effect; if the weighting is too high and the prior is unreasonable, then it takes forever for score to become close to representative.
Based upon my observations, I'd say that the prior is too high, and that the weighting is much too great. I recommend completely ignoring it except to the extent that you're interested in observing bad statistical practice or can't see the raw mean. But I don't think that the raw mean will ever be a great measure of general opinion, both because of sample-selection bias and because people don't have commensurate senses of scale. And, even if we had a good sense of general opinion, it wouldn't tell you whether you would or should like the series.
The first number is an attempt at Bayesian averaging; the second is the ordinary average of individual scores.
A Bayesian rating starts with what is called a “prior mean” or just “prior”, and a weighting; it is as if a bunch of folk have already scored the series. When the first actual scores are entered, they begin to count in the calculation; over time, the relative significance of the prior and its weighting become ever smaller.
If the prior is reasonable, and if the weighting is appropriate, then a few yahoos don't cause a series to be scored unreasonably high or low. If the weighting is too low, then the prior has almost no effect; if the weighting is too high and the prior is unreasonable, then it takes forever for score to become close to representative.
Based upon my observations, I'd say that the prior is too high, and that the weighting is much too great. I recommend completely ignoring it except to the extent that you're interested in observing bad statistical practice or can't see the raw mean. But I don't think that the raw mean will ever be a great measure of general opinion, both because of sample-selection bias and because people don't have commensurate senses of scale. And, even if we had a good sense of general opinion, it wouldn't tell you whether you would or should like the series.
Dex-chan lover
- Joined
- May 20, 2018
- Messages
- 3,426
@Oeconomist
I was referring to the rating on the official english link, which is a 4.8/5 which would be like, a 9.6 here.
I was referring to the rating on the official english link, which is a 4.8/5 which would be like, a 9.6 here.
Dex-chan lover
- Joined
- Nov 20, 2018
- Messages
- 5,157
@starburst98—
Okay. You must be a somewhat reflexively curious person to have investigated that site.
Okay. You must be a somewhat reflexively curious person to have investigated that site.
Dex-chan lover
- Joined
- Nov 20, 2018
- Messages
- 5,157
@kingofhell99—
I did a back-of-the-envelope calculation (literally using the back of an envelope). My calculation indicates that the prior is 8 and that its weighting is 62. In effect, once the first person actually scores a series, the ballot box is stuffed with 62 scores of 8.
If everyone gave a series a score of 1, it would take more than 370 actual scores for the MangaDex Bayesian score to drop below 2. (If everyone instead gave the series a score of 10, it would take 62 actual scores for the MangaDex Bayesian score to climb to 9.)
EDIT: It appears that the prior bias as been increased since the last time that I checked. Now the equivalent of 87 scores of 8 are stuffed into the ballot box along with the first real score, so that the minimal initial result is 7.92.
EDIT: The bias has been increased again. Now the equivalent of 97 scores of 8 are stuffed into the ballot box.
I did a back-of-the-envelope calculation (literally using the back of an envelope). My calculation indicates that the prior is 8 and that its weighting is 62. In effect, once the first person actually scores a series, the ballot box is stuffed with 62 scores of 8.
If everyone gave a series a score of 1, it would take more than 370 actual scores for the MangaDex Bayesian score to drop below 2. (If everyone instead gave the series a score of 10, it would take 62 actual scores for the MangaDex Bayesian score to climb to 9.)
EDIT: It appears that the prior bias as been increased since the last time that I checked. Now the equivalent of 87 scores of 8 are stuffed into the ballot box along with the first real score, so that the minimal initial result is 7.92.
EDIT: The bias has been increased again. Now the equivalent of 97 scores of 8 are stuffed into the ballot box.
- Joined
- Mar 18, 2019
- Messages
- 113
@Oeconomist @starburst98 thank you both for the clarification,i don't really like that type of rating but i can see why they have it ,but i think that Bayesian rating should be the smaller one of the two ratings(like in the smaller font)
Dex-chan lover
- Joined
- Nov 20, 2018
- Messages
- 5,157
@kingofhell99—
The presumption of those who use Bayesian scores is that they are more meaningful than raw scores until a large number of actual scores are received; so we just won't see the Bayesian score made smaller or greyed or somesuch.
There are various possible reforms, some of them preserving the present prior and weighting. The Bayesian rating could be automatically removed after one of various combinations of time-passage and actual scores. Or a more sophisticated formula could be used, so that it took no more actual scores to pull the Bayesian score to a 2 than it would to pull it to a 9.
After a large accumulation of scores across manga, the administrators might change the prior to reflect observed responses.
The presumption of those who use Bayesian scores is that they are more meaningful than raw scores until a large number of actual scores are received; so we just won't see the Bayesian score made smaller or greyed or somesuch.
There are various possible reforms, some of them preserving the present prior and weighting. The Bayesian rating could be automatically removed after one of various combinations of time-passage and actual scores. Or a more sophisticated formula could be used, so that it took no more actual scores to pull the Bayesian score to a 2 than it would to pull it to a 9.
After a large accumulation of scores across manga, the administrators might change the prior to reflect observed responses.
- Joined
- May 10, 2018
- Messages
- 39
ML is craycray🤣
Similar threads
