s********r
发帖数: 297 | 1 大家好，我是个新手，工作中刚接到一个project，是关于找是否存在correlation的。
之前的一个analyst用 proc means和proc freq总结出的报告是两个variables从图象或
者是从summarized data table看不出有relationship。现在需要我用statistical way
去找是不是correlated。我本身不是纯统计的，所以能想到的就只有 proc corr的
peason correlation看-1到1之间的correlation coefficient和p value。另外想到用
proc reg 做个regression。但是我之前都没做过类似的project，请问各位一般类似的
问题都用什么方法找correlation啊，还有run完proc还需要检查什么吗？谢谢大家的指
点，谢谢!
p.s 只有两个variable，不是mulvariate |
|
d**********0
发帖数: 222 | 2 我刚接到同样的一个活。先确定你要分析的2个变量是不是连续，或者是0/1。然后run
proc corr,看r值，再proc gplot 画图更直观显示。 |
|
s********r
发帖数: 297 | 3 谢谢楼上大家的回答，我的两个变量一个是score (1-9) 另一个是amount。谢谢，我会
用proc corr和regression都试试，然后用proc gplot画图。 |
|
z********8
发帖数: 172 | 4 看了63+50+130， 用了一个星期 sas adv pass
63题大概考了50%原题，还有6道左右是变形题， PROC SQL加了两题，efficiency加了
几题。
填空题有两道，set one; set two跟什么等价，out union corr
还有一题是 if variable= out put 请看SAS PREP GUIDE
SQL 里有几题考left join/right join 的区别
还有考except/except all的区别
ITEM 9 Value add的值变了, 题型没变
ITEM 11换了种问法
COALESCE的用法
ITEM 16 output换了问 except all or union all or append all
BUFNO 的用来做什么
Compress= NO/YES/CHAR的用法，用在哪些地方
MACRO的用法， Transpose的用法
HASH OBJECT的好处跟特点
%let 的用法考了两次，GLOBAL跟local的区别
题目不懂，好像问keep怎么用最efficiency keep stateme... 阅读全帖 |
|
s******s
发帖数: 2837 | 5 我觉得这两题是这样理解的。
第一题：D里面第二个dataset sum以后colume name变成了sum(Cost),out union corr
以后就出现了三列：id,cost,sum(Cost).A里面cost大小写确实不一样，不过sas识别
var是case insensitive,所以至少不是本题考查的点，不要在意这些细节。。。
第二题：multiple set option是依序读每个dataset的obs,读到最少obs的那个dataset
就不读了。所以只用set one;set two的话，读完第一个obs就不会继续读了，因为第二
个dataset只有一个obs。
用if _n_=1，可以读取并retain第二个dataset里面那个obs的值.PDV里面存有SumY并且
一直是36.当_n_=2的时候就没第二个dataset的事情了，只读第一个dataset的obs。所
以答案是D。 |
|
b******e
发帖数: 228 | 6 第二题解了我的疑惑，非常感谢！
第一题我还是有疑问，D里面第二个dataset sum以后colume name应该没变，因为没有
as或者是label重命名。
包子已发，再次谢谢！
corr
dataset |
|
t********6
发帖数: 43 | 7 Mixed model is only problematic when there is too little repeated
measurement, or too little subjects/observations. So I won't worry about.
If you don't have many countries, AR(1) and un-structured
corr matrix will probably have too many parameters to estimate (20yrs
parameter). You can try compound symmetry then. |
|
h*******n
发帖数: 458 | 8 谢谢几位。用VARCLUS能把 n 个variables分成几组，然后该怎么做呢？是每组内再根
据和Y的CORR选出几个variables吗？ |
|
r*******7
发帖数: 1811 | 9 我是用copula做的，先生成multi normal的同时在var-cov matrix中把corr设置成o.5
，再转成multi unif的。
网上可以搜到怎么求joint normal distribution的公式，但找不到怎么求multi unif
的而且uniform之间是dependent的。 |
|
d*********d
发帖数: 239 | 10 纯SQL 版 Test就是楼主的原始dataset.
PROC SQL;
CREATE TABLE Test1 AS
SELECT Account,Flag
FROM Test WHERE Flag = "C";QUIT;
PROC SQL;
CREATE TABLE Temp1 AS
SELECT * FROM
Test WHERE Account NOT IN (SELECT Account FROM Test1);QUIT;
PROC SQL;
CREATE TABLE Temp2 AS
SELECT *,Monotonic() AS order FROM Temp1;QUIT;
PROC SQL;
CREATE TABLE Test2 AS
SELECT Account, Flag
FROM Temp2
GROUP BY Account
HAVING order = MAX(order); QUIT;
PROC SQL;
CREATE TABLE Test_Final AS
SELECT * FROM Test1
UNION CORR
SELECT * FROM Test2;Q... 阅读全帖 |
|
M****e
发帖数: 3715 | 11 画三条线看夹角
X.Y 夹角>45o, <90o
X,Z 夹角>45o, <90o
X在中间
Y,Z 夹角>90o 负相关
corr( |
|
j*****e
发帖数: 182 | 12 what is the dimensionality of your repsonse variable?
45=10*9÷2;28=8*7÷2.
It seems that your computation of corr coeff does not conform to the
conventional approach. Please write out your parameter of interest in terms
of the distribution first.
when |
|
j*****e
发帖数: 182 | 13 Pearson correlation is not applicable here.
You have miss-interpreted the concept of correlation.
Your data is high-dimensional. Try to specify your population first. Then,
what population parameter are you trying to compare?
The way you compute corr implies that you treat each individual as a
population and genes as the random samples. Is this what you want?
The problem is complex. Please consult an expert on high-dimensional data
analysis in person. |
|
J******m
发帖数: 97 | 14 补充一点： 可以把不相关的（Irrelevant )的变量去掉， 用proc corr, 得到一个图
叫scatter plots of ranks of Spearman VS Hoeffding, 然后做 empirical
logit plot对个别可疑变量。欢迎高手指正！谢谢 |
|
|
R******d
发帖数: 1436 | 16 我不是为了筛选变量，是为了找两类数据里面可能存在的相关关系。
所以把所有相关系数都算出来先。
logistic |
|
d*********d
发帖数: 239 | 17
SAS. Proc corr might be an option if you just want to check the correlation
ship .
Or one way ANOVA. |
|
n**********0
发帖数: 66 | 18 单纯看你问的问题,,可以proc corr或者直接用regression。但是再看你的数据，为什
么一个是平均另外一个不是平均值呢呢？如果病人多了，平均看每个病人时间可能没有
太大变化，可是住院时间加起来就不一样了，还有就是每个医院医生数目也不一样，你
就看这两个似乎太过于简单了 |
|
P*****i
发帖数: 63 | 19 多谢解惑。
我现在明白了，开始只想着说一定不要有两列g3出现，指明一个就好。以为后面的full
join能把所有行cover.
现在我更加清楚coalesce这个选项的含义了---不仅是压缩同名行，而且还取全集。
自己做点笔记小结一下跟SAS的对比：
SAS里的match-merge语句，
data merged;
merge one two;
by x;
run;
效果等同于proc sql里的横向合并，具体情况又分：
当比对的by variable x全部匹配时，merge等同于inner join, 如果x只有部分值能匹配，那么等同于full outer join搭配coalesce选项；
SAS里的数据集拼接(concatenate)：
data three;
set one two;
run;
等同于proc sql里的纵向合并中的outer union corr. |
|
t**********y
发帖数: 374 | 20 I did some test runs using built-in dataset and got different outputs:
1. ===
> test=lmer(Reaction~Days+(Days|Subject),sleepstudy)
> summary(test)
Linear mixed model fit by REML ['lmerMod']
Formula: Reaction ~ Days + (Days | Subject)
Data: sleepstudy
REML criterion at convergence: 1743.6
Scaled residuals:
Min 1Q Median 3Q Max
-3.9536 -0.4634 0.0231 0.4634 5.1793
Random effects:
Groups Name Variance Std.Dev. Corr
Subject (Intercept) 612.09 24.740
... 阅读全帖 |
|
c****t
发帖数: 19049 | 21 没有。corr就是算一下linear dependence。你要找causality要上granger test这类东西 |
|
n*****3
发帖数: 1584 | 22 I think it depends on the question:
nearest neighborhood, Edit Distance , or just the the corr like last post. |
|
R*******e
发帖数: 25533 | 23 http://www.youtube.com/watch?v=k6BU6Nb_vDM
You're only just a dreamboat
Sailing in my head
You swim my secret oceans
Of coral blue and red
Your smell is incense burning
Your touch is silken yet
It reaches through my skin
And moving from within
It clutches at my breast
But it's only when I sleep
See you in my dreams
You got me spinning round and round
Turning upside-down
But I only hear you breathe
Somewhere in my sleep
Got me spinning round and round
Turning upside-down
But its only when I sleep... 阅读全帖 |
|
y***i
发帖数: 5492 | 24 来自主题: _ChenChuSheng版 - 存点东西 报告一下刚刚从WAMU wire 招行, FREE
这次用的是international wire, 而不是过去的domestic方法. 周二下午转账的,周三
国内就收到了.很快. 没有收一分钱手续费.
方法比以前的domestic简单.
Beneficiary Name: 收款人姓名(汉语拼音)
Beneficiary address: 收款人地址，e.g. Room xxx, No. xxx, xxx Road, City,
区号, China
Account number to be credited at receiving bank: 收款人在招行的账号16位一卡通
Receiving bank routing or SWIFT code: CMBCCNBS (招行的SWIFT)
Country: China
Receiving bank name and address: China Merchants Bank H.O. Shenzhen, CMB
tower #7088, Shennan Boulevard, Shenzhen, 518040, China
Corres |
|
w***y
发帖数: 2537 | 25 【 以下文字转载自 Stock 讨论区 】
发信人: mitbbs2020 (bbc), 信区: Stock
标 题: 其实GDP数据没那么重要，刚才测试了一下00年以来的数据
发信站: BBS 未名空间站 (Thu Aug 26 20:32:47 2010, 美东)
之前是GDP Annualized，08年起是GDP QoQ Annualized,
SPX跟GDP数据，GDP实际数据 Beat GDP Survey，SPX日涨，
under GDP Survey，SPX日跌，在120次数据中仅有54次。
另外66次，GDP和SPX是反关系。
我感觉SPX跟GDP报告关系不大。
Date Time Event Survey Actual Prior Revised SPX
Change % Surprise Corr
08/25/2000 08:30 GDP Annualized 2Q P 5.40% 5.30% 5.20%
8.00% -0.12 -0.10% 1
09/ |
|
h*****8
发帖数: 4754 | 26 发现我的BAC那天居然买了两份，
FIDELITY 很慢的时候我用LIMIT和MARKET下过两次单。
当晚只显示了一份，周五看居然上了两份。
整了个这个：
YOU BOUGHT CORR MARKET/BLOTTER CORRECTED CONFIRM |
|
B*****y
发帖数: 843 | 27 Addressing Mussina by his nickname, Moose, Chamberlain asked if he wanted to
join in playing video games. No, the 39-year-old Mussina told the 22-year-o
ld Chamberlain; he does not play video games.
Want to watch me play video games? Chamberlain asked. No thanks, Mussina sai
d. Well, Chamberlain continued, would Mussina like to play Ping-Pong? Withou
t answering, Mussina kept walking out the door and turned left down the corr
idor.
So Chamberlain added in a loud voice, “Hey, is the hearing the f |
|
B*****y
发帖数: 843 | 28
to
-o
sai
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
这个问题非常的孩子气, steven就喜欢问别人: want to see I play?
Withou
corr
thing
everything |
|
t*n
发帖数: 14458 | 29 The Corrs---17 videos. (I like them so much......hohoho...especially
"Forgiven not Forgotten" and "breathless")
Santana---3 videos.
the beatles---2 videos. |
|
b********t
发帖数: 992 | 30 i like the corrs, what can i do to make u love me.......
dreams..........
当当当.......... |
|
t*n
发帖数: 14458 | 31 i have The Corrs on my server. If you want I can also put
some Secrect Garden's mp3 there. |
|
t*n
发帖数: 14458 | 32 stoa---autumn
bjork---play dead
garbage---the world is not enough
the corrs---forgiven not forgotten
东方快车---让世界多一棵心 |
|
T**e
发帖数: 424 | 33 Breathless
The Corrs
Go on go on
Leave me breathless
Come on [echo...]
Hey... yeah...
The daylight's fading slowly
But time with you is standing still
I'm waiting for you only
The slightest touch and I feel weak
I cannot lie, from you I cannot hide
And I'm losing the will to try
Can't hide it (can't hide it), can't fight it (can't fight it)
So go on, go on, come on, leave me breathless
Tempt me, tease me, until I can't deny
This loving feeling (loving feeling)
Make me long for your kiss
Go on (g |
|
