a***r
发帖数: 420 | 1 exp(logsum)=exp(temp)+exp(logsum)
=exp(temp)*(1+exp(logsum-temp))
然后两边log
等式左边的logsum是这一个loop里被赋的值
右边的logsum是上一个loop中得到的值 |
|
s*****n
发帖数: 2174 | 2 问题是这个有什么用吗?
对于 1e-400这个级别的数, 取log就是-Inf了吧?
这和直接ignore也没什么区别啊? |
|
h**********e
发帖数: 44 | 3 那倒没有吧，难道不是log(1e-400)=-400log10？ |
|
l*********s
发帖数: 5409 | 4 1st order Taylor expansion of log function. |
|
a***r
发帖数: 420 | 5 啊，是我原帖没说清楚
其实temp=function1（i）*function2（i）
两个都小到1e-200左右这种级别，乘起来就被ignore了
log之后基本能都被挽救 |
|
h******3
发帖数: 190 | 6 I looked at the hypergeometric test statistics for log-rank test. It seems
that the test won't work if the two survival curves cross each other in the
middle of the testing period.
But I am not sure, since I just started to pick up biostatistics. Thanks! |
|
d******g
发帖数: 130 | 7 I think the large sample will make the test with -2log(likelihood ratio)
more valid. Since we take the log and double it to yield an approximate chi
-squared sampling distribution and we use LARGE sample to approximate this
chi-square distribution.
Also when H0 is false, the large the sample size, the larger the -2log(likelihood ratio). |
|
s*r
发帖数: 2757 | 8 或者什么editor有合适的syntax file
可以让log file显示得和在PC SAS里运行完之后显示的一样的效果 (note 行是蓝色，
error行是红色)? |
|
s*r
发帖数: 2757 | 9 这不得看error的上下文吗
很多syntax highlight都是针对关键词的，而非整行的
怎么自己写？ 用perl在把log文件转成html的过程中着色? |
|
c****g
发帖数: 156 | 10 There are lots of SAS log check macros, just Google it. |
|
D******n
发帖数: 2836 | 11 syn match sasNote "\s*NOTE:.*"
syn match sasWarn "\s*WARNING:.*"
syn match sasError "\s*ERROR.*"
syn region sasEp start="^\s\+_$" end="^\s\+\d\+$"
hi sNote term=NONE cterm=bold ctermfg=Blue ctermbg=Black gui=bold
guifg=Blue guibg=black
hi sWarn term=NONE cterm=bold ctermfg=Green ctermbg=Black
gui=bold guifg=Green guibg=black
hi sError term=NONE cterm=bold ctermfg=Red ctermbg=Black gui=bold
guifg=Red guibg=black
hi def link sasNote sNote
hi def link sasWarn sWarn
hi d... 阅读全帖 |
|
D******n
发帖数: 2836 | 12 If you don't have perl installed, this won't help you.
#!/usr/bin/perl -w
$0=~/.*\/(.*)/;
use Term::ANSIColor;
die "\nUsage: $1 \n\n" if @ARGV!=1;
my @content = `cat $ARGV[0]`;
for (@content)
{
if ($_=~/NOTE:/)
{ print color 'bold blue';}
elsif ($_=~/WARNING:/)
{ print color 'bold yellow';}
elsif ($_=~/ERROR/)
{ print color 'bold red';}
print $_;
print color 'reset';
} |
|
s*r
发帖数: 2757 | 13 似乎还depends on cygwin
如果log file所在的目录有空格的话，cgywin的msg会跳出来
dos window下没看到彩色，只有这样的字，不知道是哪里没设好
←[0m←[1;34mNOTE: DATA statement used (Total process time):
←[0m real time 0.61 seconds
←[0m cpu time 0.58 seconds |
|
h******3
发帖数: 190 | 14 I found the following online, but realize "loglogs=logh" that is supposed to
give log cumulative hazard output does not work...
"proc phreg data='c:\alda\rearrest' noprint;
strata personal;
model months*censor(1)= /ties=efron;
baseline out=strata covariates=null logsurv=lsurv loglogs=logh /method=pl
nomean;
run;
"
??
thanks for help~ |
|
s********u
发帖数: 344 | 15 比较两个样本的均值。样本说大不大，说小不小， 1000左右，完全不是正态分布。因
为样本还算大，能够用上central limit theorem 吧。那就z检验吧，得到两个样本均
值是不同的结论 p=0.014。
没事有试了一下log transformation，看上去正态多了，结果得到了p=0.24。
也看过有些文献说做了transformation很多统计量会变，但是这样的结果我该怎么解释
呐？ |
|
s*r
发帖数: 2757 | 16 log transformation actually checks whether the ratio of 2 geometric mean is
1. i guess you will get similar p if you use permutation test to get an
empirical null |
|
r********2
发帖数: 19 | 17 既然用了clt，我们应该不需要原分布是normal的吧。
这个是应该可以预计到的结果。log transformation 对不同值转换的效果不一样，在
>1的range 里，使得两个数的距离变小。所以变得不显著了啊。但是两个test是在检验
不同的mean。
(司徒丢丢) 的大作中提到: 】 |
|
j*****e
发帖数: 182 | 18 Based on what you said "没事有试了一下log transformation，看上去正态多了，结
果得到了p=0.24。"
log_normal can be used to model the two distributions.
Let's say the first distribution is log_normal (mu1, sigma1), the second
distribution is log_normal(mu2,sigma2).
Since your p-value is large (0.24), mu1 is not sig. different from mu2.
The mean of the fist log_normal is exp(mu1+0.5*sigma1**2). The mean of the
second log_normal is exp(mu2+0.5*sigma**2).
You said "那就z检验吧，得到两个样本均值是不同的结论 p=0.014。
This implies sigma1 is not ... 阅读全帖 |
|
c*******g
发帖数: 695 | 19 有一些PK 的变量比如AUC CL Cmax之类的
计算平均值都是做log transform之后来算
或者算geometric mean
请问这样的变量的SD应该怎么算？
谢谢 |
|
h********o
发帖数: 103 | 20 You can use array like this:
==================================
data test;
input var1 var3 var2 var10 var6;
array temp(5) var1 -- var6;
array logvar(5);
do i = 1 to dim(temp);
logvar(i) = log(temp(i));
end;
drop i var1 -- var6;
cards;
1 2 3 4 5
; |
|
p***r
发帖数: 920 | 21 or you can do it in another brutal way
*WIDE TO LONG;
PROC TRANSPOSE DATA=data1 OUT=data2;
BY var_id;
VAR _ALL_;
RUN;
data data3;
set data2;
log_var=log(col1);
run;
*LONG TO WIDE;
PROC TRANSPOSE DATA=data3 OUT=data4
BY var_ID;
ID variable;
VAR log_var;
RUN; |
|
m*n
发帖数: 695 | 22 有三组数据， 需要做两两比较， 但不符合one –way ANOVA 的要求，log 转化后仍然
非正太。
我用的是spss, 做了non-parametric K-S 分析， p< 0.05, 该如何再做两两比较呢？
看文章中有人又用了ANOVA 下的post hoc test 中方差不齐的一个test 来做
multiple comparison. 这样可以吗？
还有人用 non-parametric K-S 单独做了两两的比较， 共做了三次。 这样做p 是不
是要很低才行？
向这种数据该如何做多个样本的两两比较啊？
请高手指点。
万分感谢！！！ |
|
p********6
发帖数: 1339 | 23 只要分布不是特别skewed，数据量不是很小，可以用ANOVA。
一次log transformation之后如果还是很skewed，可以试试Box-Cox transformation。 |
|
a*****i
发帖数: 1045 | 24 同样的疑问，做了log transformation的意义何在呢?还是只是单纯的为了分析数据方
便呢？ |
|
p***r
发帖数: 920 | 25 比如化学中溶液中离子的浓度，做log transformation 转化成类似于pH值一样的量分
析起来就容易一些，是实际需要。 |
|
a*****i
发帖数: 1045 | 26 同样的疑问，做了log transformation的意义何在呢?还是只是单纯的为了分析数据方
便呢？ |
|
p***r
发帖数: 920 | 27 比如化学中溶液中离子的浓度，做log transformation 转化成类似于pH值一样的量分
析起来就容易一些，是实际需要。 |
|
p********r
发帖数: 1465 | 28 看是对自变量还是因变量还是都做log，不同情况下解释不同 |
|
l**k
发帖数: 419 | 29 我没有收到你的信呀.
但是我不能把log给你, 那里面涉及到太多的用户隐私了. |
|
a********n
发帖数: 102 | 30 每次都要log in 太麻烦了，我一天要键入几十次。
多谢！ |
|
e*****c
发帖数: 37 | 31 It happend to me twice, when I wrote long articles under WWW, the system didn't consider I was active and logged me out automatically. The worst thing was the article got lost, I only found it out after I submitted time consuming articles and had no way to get my articles back! A better solution would be when the submission failed, system returns to the composer with the original article opened in that window. |
|
s*******n
发帖数: 3 | 32 I cannot log on using my previous username: smxzr
(I was told that the password is WRONG - but it is the correct one)
pls help me.
I have to register a new username to report my issue.
thanks. |
|
m****s
发帖数: 18160 | 33 这个文章版面是指的哪块？
刚分别测试了分类讨论区进去和我的讨论区进去版面，并没有log out。
如果再有发生，麻烦注意一下还有没有别的特别之处。 |
|
m********o
发帖数: 975 | 34 补写星期二的LOG
跑步的日子, 本来应该是去TRACK跑INTERVAL的,没有能够按照计划成行. 只能在跑步机上了练练了.
ELEVATION 2
HIGH SPEED 6.8
LOW SPEED 5.2
DURATION: 20分钟 |
|
j*********g
发帖数: 3826 | 35 log 很勤快，练得很勤快，下个礼拜我也得恢复力量练习了。 |
|
m********o
发帖数: 975 | 36 明天出门了, 下个星期我就没LOG了, 争取在艰苦条件下创造条件, 能做多少就做多少
吧. |
|
m********o
发帖数: 975 | 37 我好好的一个健身LOG啊, 居然给你们搞成这个样....
每人给我一个包子, 我以后就啥也不说了. |
|
c******t
发帖数: 1733 | 38 灌水太花时间，今天开始把网戒了。等会我的帐号我自己也log 不进去了， 有什么健
身方面的问题，请发到我的gmail信箱交流。 |
|
z*****a
发帖数: 9790 | 39 妹子为啥不去fitness帖log？有包子发。。。。。 |
|
|
k***z
发帖数: 4718 | 41 11/08/09: Table Tennis 2hrs. Just remember I lost all my training log on
iphone, sigh.. |
|
w*******y
发帖数: 60932 | 42 This deal is shipped and zero tax. I sent this a gift to my parents and also
got some stuff shipped to me. Total with multiple shipping addresses was $
19.09!
Go to Swiss Colony website here
Look up Postpaid Holiday Bars & Logs in the search bar. Price is $19.95.
Decide who to send it to and add to cart. Since postage is included, this
really is a super good way to get the free chocolates and/or nuts!
Now the fun part! Decide which two freebies you want and enter the code into
the search bar:
FR... 阅读全帖 |
|
k*****e
发帖数: 93 | 43 第一次有android device,是tablet.
用google play的时候注册了自己的gmail账户。
现在的问题是怎么log out? 就像iphone sign out一样。
我一直没找到。
请高手指教，谢谢。 |
|
k*****e
发帖数: 93 | 44 第一次有android device,是tablet.
用google play的时候注册了自己的gmail账户。
现在的问题是怎么log out? 就像iphone sign out一样。
我一直没找到。
请高手指教，谢谢。 |
|
l*****r
发帖数: 2123 | 45 I use pokerstars and I can log in in China |
|
j******s
发帖数: 48 | 46 好吧，自己回答一下思路，虚心求各路大神拍，给点意见
第一题应该是建立一个binary tree,然后recursive求每一个节点的左右子树的最高高
度,他们和就是在这一位上的largest distance，可以online做，但是需要在每个节点
保存左右子树的高度并且动态更新。
o(n) time + O(n) space
第二题,
Assumption:
1 Log file is typically very small, operations are append, delete and read.
2 Cluster is built on Hadoop, which has a chunk size of 64MB, roughly, and
it might be too large and inefficient if we use this chunk size for the log
file.
3 Log information is typically not very important, less effort is needed for
redundan... 阅读全帖 |
|
l*****9
发帖数: 9501 | 47 Your input can't be opened:
VLC is unable to open the MRL '1600.m3u8'. Check the log for details.
Your input can't be opened:
VLC is unable to open the MRL '2400.m3u8'. Check the log for details.
Your input can't be opened:
VLC is unable to open the MRL '3000.m3u8'. Check the log for details.
Your input can't be opened:
VLC is unable to open the MRL '4500.m3u8'. Check the log for details.
Your input can't be opened:
VLC is unable to open the MRL 'duh.m3u8'. Check the log for details.
Your input ... 阅读全帖 |
|
|
