|
|
b********s 发帖数: 6928 | 3 dude it would be difficult to explain what regression is in a few sent
ences. basically it helps you explore how Y changes with X changing. I
f you have more than one X, then it helps you explore how Y changes wi
th X1 changing and all the other X held constant
anyway, if you don't need the exact relationship between the two varia
bles, just use any available point and click software, run descriptive
statistics, and look for the correlation. if the value is larger than
, say 0.6, it is strong re... 阅读全帖 |
|
b********s 发帖数: 6928 | 4 well then it seems like a linear relationship with a limited range of
the independent variable, i.e. the power |
|
b********s 发帖数: 6928 | 5 R^2 explains how much of the variance in Y is captured by X
if its value is 1, means that 100% of the variance in Y is captured by X
if you don't run the regression, don't worry about this one |
|
p********l 发帖数: 32 | 6 excel都可以做啊....
选数据分析.
你google下好了... |
|
p********l 发帖数: 32 | 7 是不是线形最简单方法是先描点看啊...
然后run完看看R-sqr好了 |
|
|
p********l 发帖数: 32 | 9 附带,你到底只是要知道两个变量相关性(那叫correlation), 还是要知道两个变量的具
体等式阿(比方,一个确确的方程告诉你,y自变量是如何由x因变量确定)? |
|
p********l 发帖数: 32 | 10 excel? version 2007 is build in.
just click the office image (left-top corner), then click "excel option",
then click "adds in". u will find it there.
old version excel prob. needs adds in...forget about that, but should be
easy to download online if that is the case...
suddenly i can not type Chinese...-____-
GL |
|
p********l 发帖数: 32 | 11 btw, if u only want to see the correlation, a simple function in excel calle
d "correl" does this job~. |
|
b********s 发帖数: 6928 | 12 找到了,多谢啊!我原来都不知道excel还可以做这些…… |
|
p********l 发帖数: 32 | 13 you are very welcome, my pleasure ^_^ |
|
h*******x 发帖数: 12808 | 14 谢谢微风,微风好专业啊。
我最近一直没有时间来学R |
|
|
h*******x 发帖数: 12808 | 16 就看到orrelation的程度把
不用具体的model |
|
|
p********l 发帖数: 32 | 18 GOOD NEWS FOR YOU.
you can use online google excel.
and it luckily does have the "correl" function build in, so just paste your
data, and use the "correl" function~~
GL |
|
p********l 发帖数: 32 | 19 click your mouse into any cell, then type "=correl(", then use your mouse to
click the first series data, then type",", then use your mouse to select yo
ur scond series of data, then tyep ")", done~
your |
|
|
h*******x 发帖数: 12808 | 21 我的数据两比较大,大概有几十万个。
excel之类只能处理小于65k个数据吧。我这里有数十个文件,很多文件的数据都超过几
十万。excel之类不是很给力。
your |
|
h*******x 发帖数: 12808 | 22 微风,在R里面怎么算相关性啊?能用什么函数啊?或者什么package? |
|
b********s 发帖数: 6928 | 23 我不精通的,用 cor( var1, var2)试试吧,default是Pearson's product-mome
nt correlation coefficient,是最基本的function |
|
p********l 发帖数: 32 | 24 是啊. 老excel处理60k+行...不好意思不知道你多少数据.
csv或者excel2007可以处理1000k+行,而且可以使用correl 方程.
最后..你随便是用一个统计软件(比如:sas/gauss/stat/eview,或者随便搞个编成软件写
个方程就可以阿,比如C++)
....这个问题折磨你好久阿, GL!! |
|
b********s 发帖数: 6928 | 25 类似这样的东西你难道不是大侠!
setwd("DIRECTORY")
data = read.csv("DATA.csv")
Y=data[,1]
X=data[,2]
correlation=cor(var1, var2)
summary(correlation) |
|
b********s 发帖数: 6928 | 26 那些软件都要钱的吧。。。话说他cs的编程大侠,竟然要问怎么用R,我觉得他是
故意耍咱们呢,哼哼
件写 |
|
p********l 发帖数: 32 | 27 a a a...被调戏了...
俺闭嘴了... |
|
b********s 发帖数: 6928 | 28 别,他估计是真不会做,因为他不知道啥是correlation,你帮帮他吧 |
|
h*******x 发帖数: 12808 | 29 好的,谢谢啊,我去查查。
var1,var2是数组吗? |
|
|
h*******x 发帖数: 12808 | 31 微风太牛了,太感谢啦!!
我完全不会统计的东西啊 |
|
|
|
|
|
b******a 发帖数: 12216 | 36 我就知道结果是什么意思,中间怎么计算的我咋会知道涅。。。
你去wiki啊,一般都会告诉你这个到底怎么计算的。 |
|
b********s 发帖数: 6928 | 37 var1, var2是csv里面的column lable,就是variable name,呵呵 |
|
d**********u 发帖数: 4124 | 38 谢谢这个帖子里校友们对潇妹的帮助~赞~
每人一个包子:) |
|
|
B******u 发帖数: 23763 | 40 老弟阿,你这个人,唉,俺咋说你呢?
你躲一边看看也就是了,
插啥嘴阿。 |
|
|
|
|
s****e 发帖数: 3310 | 44 笑死俺了,这话说的好像你别的会的很多似的。
这是大一水平的题好不好,P大真是一辈不如一辈了。 |
|
|
|
d**********u 发帖数: 4124 | 47 赞潇妹大方~
啊,这样啊,那我给版里面捐点包子把。 |
|
|
s****e 发帖数: 3310 | 49 俺不会啊,怎么帮。俺虽然没帮上忙,在主观意愿上也是支持潇妹的吧,也算精神可嘉
对不
小童鞋你帮潇妹了么?就看到你捣乱了.... |
|
|