vegetable8

回复 15# not4weak

数学家

大胖熊

not4weak 都是教授级别的人啦，线性回归，曲线拟合，样条插值，这些都是大三
就学的。以后大家有问题，尽管问他。。

qzhou3

回复 1# puccafly

我用英文解释一遍
What you have is a linear regression. For linear regression, the R2 ranges from 0 to 1. If it is 0, the two variables, X and Y, are not associated. If R2 is 1, the two variables are perfectly associated. In real life, variables are rarely perfectly associated. Thus R2 is rarely 1.
The lower case r, called correlation coefficient,  is the correlation between two variables. It ranges from -1 to +1, where 1 is total positive correlation, 0 is no correlation, and −1 is total negative correlation. The upper case R is the same as r if there is only one independent variable (predictor).
R2 is called coefficient of determination. It tells how much variance in Y can be explained by X. In your case, how much variance in living space is explained by income.
In your example, you use X, the income, to predict Y, the living space in square meters. Your model can predict 97.67% of the variance in living space. This is a super strong association, almost too good to be true...
When you use excel or statistical software to perform regression analysis, the calculation is based on least squares method. That is, the sum of squares of the distance from your data points to the regression line is the smallest. Hence your points are rarely sitting on the line; when a point is on the line, the error for that point is 0.
Your orange line is not a formula. It is just a connection of the points.

vegetable8

又是一个数学家或统计学家

vegetable8

回复 19# vegetable8

我是指qzhou3

not4weak

回复  puccafly

我用英文解释一遍
What you have is a linear regression. For linear regression, th ...
qzhou3 发表于 2014-10-7 00:45

不对，那条orange line 是一个函数， 不是连线

qzhou3

回复 21# not4weak

quadratic function?

not4weak

回复 22# qzhou3

你也是对的， 她这数据比较规则，在eXcel 下可以加trendline 公式就出来了。 polynomial 也可 我试一下

not4weak

11.059ln(x) -78.068

Wow! 这么规则  r^2 =1.0

估计是伪造的吧？

qzhou3

回复 24# not4weak

Too good to be true :)   Look like fake data

puccafly

"只要在连接处的7 个点保持一阶导数和二阶导数连续就可以了"

所以最终可求得一个那个橙色曲线的函数方程？不是6个。。
艾玛，这种时候就无比遗憾没有工科南喷有，一个会算样条插值的

puccafly

11.059ln(x) -78.068

Wow! 这么规则  r^2 =1.0

估计是伪造的吧？
not4weak 发表于 2014-10-7 10:45

中国指数研究院的网上文件中当下的数据，图二的那组画出来完全就是个直线，图一的也没好多少阿这样看来

中指院阿 ，名字看起来很正式，怎么实力跟文献这么坑爹？！

not4weak

"只要在连接处的7 个点保持一阶导数和二阶导数连续就可以了"

所以最终可求得一个那个橙色曲线的函数方程 ...
puccafly 发表于 2014-10-7 11:14

理科南，哈哈，彻底明白了吧

puccafly

理科南，哈哈，彻底明白了吧
not4weak 发表于 2014-10-7 12:08

我一直以为你的头像是条飘逸的红色纱裙。。。