g*****y
发帖数: 7271 | 1 所以不能太晚用。比方我觉得一个比较方法是；
1. 假设至少打4分，领先2分或以上的赢
2. Baseline：固定第3个球用
3. 改进如下：如果第2球之前的分数是15：0，提前用。如果不是，
则等着第3球再用。这个应该能比baseline好吧？ |
|
g*****y
发帖数: 7271 | 2 你是模拟派，我还是直接推算一下吧。我试着改成3局2胜来证明一下。
其它一律按照Snape原来的假设。
假设A/B打3局，best of 3. A可以使用一次制胜武器。看怎么
优化制胜武器的使用时机来提高A的胜率：
Baseline：如果没有所谓的BP的话，我们设计A上来第一局
第一分就用制胜武器。第一局A胜的概率为7/15
后面两局没有制胜武器可用了。所以每局赢率为1/5
A的胜率是：7/15 * （1-4/5 * 4/5） + 8/15 * 1/5 * 1/5 = 0.189
BP optimized version：如果第一局 1：0领先，则用。
否则，留到第二局第一分使用（应该还可以优化，懒得做了）
第一局有两种可能：
1. 1/3的概率，1：0领先，用制胜武器，胜第一局
2. 2/3的概率，0：1落后，留制胜武器到第二局第一分。
Case 1：A胜第一局，然后再拿至少一局的概率是
1 - 4/5 * 4/5 = 9/25
Case 2：
A第一局保留制胜武器，仍然获胜的可能，只有1/15
A第二局直接使用制胜武器：胜率7/15
A第三局胜率 1/5
A拿到至少两局的概率是：7/... 阅读全帖 |
|
G***a
发帖数: 27294 | 3 You cannot do this in the long run.
As we grow older, we lose muscle mass. The more we lose, the lower
metabolism we will have.
With extreme diet, baseline metabolism will decrease even more.
For example, you may burn 1,600 calories per day without exercise now.
With simple calorie control and no exercise, you may only burn 1,450 ten
years later.
Then you will reach a point, no matter how hard u try, how little you eat,
you stay on plateau.
That' why simple extreme diet control works for young g... 阅读全帖 |
|
D*******a
发帖数: 1723 | 4 不明白你为什么撒谎呢，明明是"89% had variable weakness of lower extremity
muscle group at 1 month.:翻译成中文是"第一个月时有不同程度的下肢肌无力"
到了你嘴里，居然成了"下肢瘫痪"
At 1 year 7 patients (78%) had a reproducible increase in bladder pressure
with stimulation of the dermatome. Two patients were able to stop
catheterization and all safely stopped antimuscarinics. No patient achieved
complete urinary continence. The majority of subjects reported improved
bowel function. One patient was continent of stool at baseline and 4 were
continent ... 阅读全帖 |
|
l*****o
发帖数: 19235 | 5 随便找了个作者提供的链接看了一下。能看懂点英文的人，应该都知道作者在干吗了。
原文：
直到2010年8月在《泌尿学杂志》正式发表的其初步结果论文中，他们才承认9例脊
柱裂一年后“无一能完全自主控尿（Nopatient achieved complete
urinarycontinence）”。该结果遭到同行专家严厉批评和质疑。
原始链接：
http://www.ncbi.nlm.nih.gov/pubmed/20639040
RESULTS:
At 1 year 7 patients (78%) had a reproducible increase in bladder pressure
with stimulation of the dermatome. Two patients were able to stop
catheterization and all safely stopped antimuscarinics. No patient achieved
complete urinary continence. The majority of subjects repor... 阅读全帖 |
|
t******n
发帖数: 2939 | 6 ☆─────────────────────────────────────☆
nicolson (nicolson) 于 (Sat Mar 16 17:34:09 2013, 美东) 提到:
基本上如下：
SS现在最多1750USD，但是alex最近10年工资有一阵没有那么高，按照2004年及以前都
是8万算。2005=83000,2006=86000,2007=90000,
2008=93000,2009=96000,2010=10000,
2011=103000,1012=106000。
计算器算出来是baseline=1234美刀。
http://www.ssa.gov/retire2/AnypiaApplet.html
下面链接说全家最多收入不会超过150%of baseline,
http://www.ssa.gov/survivorplan/onyourown5.htm
所以一共1851美刀。另外他们的房子在2004年是倾家荡产20%down。40万的房子，还在
付30year fixed @3.5%,月供是1400，房产税大概400，所以每个月刚刚可以拿SSN... 阅读全帖 |
|
c*********l
发帖数: 3438 | 7 【 以下文字转载自 Military 讨论区 】
发信人: continental (飞), 信区: Military
标 题: 大数据显示：亚男白女婚姻离婚率最高
发信站: BBS 未名空间站 (Mon Mar 23 22:46:21 2015, 美东)
亚裔互相结婚的，比白人互相结婚的离婚率低55%。
白人男和亚裔女结婚的，离婚率和白人互相结婚的基本一样，略高4%，所以白人男的有
incentive和亚裔女结婚，或者说没有disincentive。
但是亚裔男和白人女结婚的离婚率比白人互相结婚离婚率还要高59%，所以亚裔男和白
人女结婚的incentive很差。
至于为什么离婚，那小3肯定是一个很大原因，虽然不是唯一原因。基本上统计来说，
白人互相结婚，肯定小3比亚裔之间要高，statistically speaking.
http://onlinelibrary.wiley.com/doi/10.1111/j.1741-3729.2008.004
第168页，table 4， full model.
coefficient算法是，take exp(.) - 1, 就是亚裔互相结... 阅读全帖 |
|
c*********l
发帖数: 3438 | 8 【 以下文字转载自 Military 讨论区 】
发信人: continental (飞), 信区: Military
标 题: 大数据显示：亚男白女婚姻离婚率最高
发信站: BBS 未名空间站 (Mon Mar 23 22:46:21 2015, 美东)
亚裔互相结婚的，比白人互相结婚的离婚率低55%。
白人男和亚裔女结婚的，离婚率和白人互相结婚的基本一样，略高4%，所以白人男的有
incentive和亚裔女结婚，或者说没有disincentive。
但是亚裔男和白人女结婚的离婚率比白人互相结婚离婚率还要高59%，所以亚裔男和白
人女结婚的incentive很差。
至于为什么离婚，那小3肯定是一个很大原因，虽然不是唯一原因。基本上统计来说，
白人互相结婚，肯定小3比亚裔之间要高，statistically speaking.
http://onlinelibrary.wiley.com/doi/10.1111/j.1741-3729.2008.004
第168页，table 4， full model.
coefficient算法是，take exp(.) - 1, 就是亚裔互相结... 阅读全帖 |
|
t********l
发帖数: 996 | 9 T好比网球中的NET PLAYER，始终站在网前，善长进攻，扣杀凶猛，是得分主力。
P好比网球中的BASELINE PLAYER， 打底线，善长防守，为网前的T创造机会得分。
两人技术都不全面但如果配合的好，双打还是没有问题的。
但是如果一旦敌人进攻厉害，P防守OFF THE COURT， T又无法跑回底线，进行防守，或
者T有点受伤，网前进攻无力，P又只能指望T得分，自己不会上网进攻，那会输的很惨。
网球中最吃香的是H， ALL COURT PLAYER， 能守能攻，能上能下，能前能后，满场跑
，技术全面， 和NET PLAYER 能搭配， 和BASELINER 也能搭配，如果两个都是ALL
COURT PLAYERS， 那就是强强连手，战无不胜。
次篇只是娱乐大众，没有任何对TPH的实际褒贬只说。 |
|
t******g
发帖数: 3641 | 10 go petplan上提到了waiting period，这个我懂，大概是说买了多久才开始生效，否则
立马得病的，算pre-existing.但是后面又有一个30天内的vet last visit，我就糊涂
了。要是我30天内不带猫去检查，难道保险就永远失效了？
* Waiting periods
As is standard in the industry, there is a waiting period on Petplan
policies before accidents and illnesses can be considered coverable. Your
pet’s accident or injury coverage will begin twenty-four (24) hours after
the policy becomes effective. There is a fourteen (14) day waiting period
after the effective date of the policy during which we will... 阅读全帖 |
|
a*****y
发帖数: 33185 | 11 什么叫高？首先要有个baseline才能判断高和不高。
但是这个baseline并不是现成的。 |
|
z***y
发帖数: 7151 | 12 这是senior 或者是architecture的位置吧。
回答这种问题， 要站在技术层面之上考虑。
我可以给你开个头， 你自己想想：
1. Set up baseline and goal
Before you taking the position, you need know what is the current status of
the business process, IT infrastructure, the database eco systems. That's
the baseline. Then you project a goal--where you want to bring them? You
also provide metrics to score your goal.
2. Have a big picture
What would you do if you were key driver on a new Disaster Recovery project?
Are you going to present several DR so |
|
r****y
发帖数: 26819 | 13 BYTE UNIX Benchmarks (Version 5.1.3)
System: cXXXX-XXXX.cloudatcost.com: GNU/Linux
OS: GNU/Linux -- 2.6.32-358.18.1.el6.x86_64 -- #1 SMP Wed Aug 28 17:19:38
UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU L5520 @ 2.27GHz (4533.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEX
IT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU L5520 @ 2.27GHz (4533.5 bogomips)
Hy... 阅读全帖 |
|
p****p
发帖数: 3360 | 14 其实你最后的两组稀释才是线性的，是比较好的结果。
如果你的primer比较好，每轮能达到90％以上的扩增率的话，10倍的稀释应该增加Ct大
约3.3－3.6之间。楼上的说你模板浓度太高应该没有错。不过Ct值小于24的都不在线性
范围内是有点奇怪(比如E-4应该Ct~21，0.001稀释应该17－18左右。其它更高的模板浓
度就不要用了。）。
4-5个cycle就出来信号，你没有办法定baseline，估计连自动baseline都很有问题。 |
|
t******o
发帖数: 7 | 15 你没有贴你的扩增曲线，我觉得你的模板浓度太高了，可以看看你的扩增曲线，估计前
几个都是锯齿形的，这样影响自动生成的baseline，可以先看看扩增曲线，看从哪个浓
度开始才是光滑的s曲线，然后把前几个样品圈掉，baseline改了估计ct会有一定改变
。估计你之后还要然后按需要做稀释，比如找1：5这样的稀释度。样本很多时候也是应
该稀释的，如果还有很低ct的估计也是太浓了。 |
|
t******o
发帖数: 7 | 16 但看样本数值，比如第一个平行管相差很大，我感觉两个5的是baseline的取值没有划
在对数增长期造成的，你可以手动提高baseline的值再看看。 |
|
A******y
发帖数: 2041 | 17 This is how I do it. You inject cells with or without matrix gel. You want
to make sure the xenograft establishes itself and start treatment at around
100mm3 (entering log growth phase). That's how you establish the baseline
value and day 0 for each mouse. The experiment can be staggered for mice
within the same group. I just love when people tell me that they started
treatment right after cell injection without establishing the tumor baseline
because I can just reject the paper based on ba... 阅读全帖 |
|
s**********8
发帖数: 25265 | 18 baseline for IC is hard to stablize. need run hrs to overnight to reach a
stable baseline. suppressor may get better after being used for a while.
monitor total conductivity during the run |
|
P***e
发帖数: 804 | 19 mobile phase B 10-30% gradient内 baseline linearly increase. 听起来像是
ACN的问题。Try higher wavelength such as 254 nm, if no baseline drifting,
most likely it is the ACN impuritiy problems.有没有试过不同batch或vendor 的
ACN？
题，但
30 |
|
p****y
发帖数: 23737 | 20 仅供参考，祝LZ和宝宝都安好！记得要保持心情平静愉快，情绪对身体的影响很大。
本文作者: 邱雪婷 营养师
http://www.twvns.org/articles/research/16-2008-07-31-09-43-31
许多研究已经證实饮食对糖尿病人血糖，血脂肪及体重等控制有莫大的影响! 而营养师
与糖尿病卫教师，也总不忘一再叮咛且教导病友如何作醣类计算，与糖份斤斤计较!
2006年发表於糖尿照护(Diabetes Care)，比较全素低脂饮食计画与美国糖尿病学会所
推广的醣类计算饮食，发现，两种皆能够有效控制血糖，然而全素饮食对血糖的降低比
糖类计算来的更加有效。作者 Barnard et al. 表示，全素饮食有效控制血糖主要原因
可能来自较低的热量及脂肪，在此情况下，全素试验组的体重降低较多，可能因此增加
胰岛素敏感度。 此外，高脂饮食，在之前的研究发现，可能造成肌肉细胞及肝脏细胞
囤积较多的脂肪，导致细胞对胰岛素不敏感! 而胰岛素不敏感是导致第二型糖尿病重
要因子。
提高胰岛素敏感度最好的方法: 运动，减重。
此研究中，营养师卫教全素组受试者，选择低脂食物，在醣类方... 阅读全帖 |
|
s****e
发帖数: 2934 | 21 ☆─────────────────────────────────────☆
Synthon (合成子·小冥王) 于 (Wed Oct 18 01:30:52 2006) 提到:
要精确的还是近似的？
近似的比较好弄，你先画个baseline，baseline以下的东西都不要了。然后画个
amorphous scattering的broad peak，剩下的就是crystalline part了。两个面积相
除就好了。
精确的比较麻烦，一堆公式我在这儿敲不出来，建议你找本polymer crystallography
的书看看好了。不过要想精确，experimentally你要把2theta角做到60度以上。。。。。
☆─────────────────────────────────────☆
possible (yes) 于 (Thu Oct 19 12:30:16 2006) 提到:
昨天online的一片大分子就是你的教材。
☆─────────────────────────────────────☆
Synthon (合成子·小冥王) 于 |
|
s**********t
发帖数: 217 | 22 Just came back from the NYC CMG seminar, seeing a lot of old friends and
meeting a lot of new friends. Still feel very excited. Many thanks to KDs
organized this wonderful meeting, letting CMGs having this platform to share
information.
As a guy happened started my clinic career couple years earlier, I would
like to share some experience with my young friends. I hope this is going to
help you go through your transition period smoothly.
When an intern reports to me A NEW ADMISSION, what do I wan... 阅读全帖 |
|
g****i
发帖数: 2269 | 23 这种低血压我见的很多，因为在chf floor。我们的病人只要asymptomatic，都不call
doc的，除非sbp 60了。当然还要比较baseline，如果baseline 高于100，现在80，还
是要小心。我们的chf病人80的继续给beta-blocker。有一次70多，我实在不想给beta-
blocker了，就问医生A，他说只要病人不pass out就继续给。
有时也取决于医生。有个病人sick sinus syndrome.第二天要上pace maker， 他的hr
29，但是他asymptomatic，胆大的医生B两次都说没事。结果害的我们的tele的纸都打
完了，我不得不叫tele tech把perimeter设置到25。结果有一人C来make round，顺便
看到这个病人hr 28，就大叫，这是谁的病人啊，怎么还不给atropine. 那个病人的医
生B很胆大，他的beta-blocker的perimeter总是：hold if hr<45. |
|
l******k
发帖数: 27533 | 24 我这个高度近视怎么没看到这两项，谢谢提醒。
HOWEVER, the most accurate way to predict renal function is 24 hours urine
output. A snapshot SCr and BUN may not reflect renal function unless we know
the baseline SCr. For example, if the pt's baseline Scr is 0.7. SCr of 1 is
still within the normal range but the pt may experience AKI. An absolute
SCr value usually doesn't give much information in the renal function. That'
s why I am not challenging MD's diagnosis by the most recent lab values.
Besides, it's very common f... 阅读全帖 |
|
R*k
发帖数: 264 | 25 http://arxiv.org/abs/1109.4897
这个网页现在很难打开，估计很多人在看，呵呵
The OPERA neutrino experiment at the underground Gran Sasso Laboratory has m
easured the
velocity of neutrinos from the CERN CNGS beam over a baseline of about 730 k
m with much higher
accuracy than previous studies conducted with accelerator neutrinos. The mea
surement is based on highstatistics
data taken by OPERA in the years 2009, 2010 and 2011. Dedicated upgrades of
the CNGS
timing system and of the OPERA detector, as well as a high preci... 阅读全帖 |
|
s*******r
发帖数: 6 | 26 第一轮考两个题，过了。第二轮开始套你的策略，要求你比一个baseline做得好。估计
那个baseline就是他们现在运用的策略。三天时间，让你一个人就用matlab试出一个更
好的策略（其实就是低风险高收益高流动性低换手率的稳定赚钱策略）。他妈的我要是
有这种策略，还用申请你们公司？ |
|
p******e
发帖数: 58 | 27 Anyone has some inside information about the recent result
on K2K long baseline neutrino Oscillation Experiment?
I heard some rumer that they have significant lower rate
than expected, is it true? This should be very important
results.
For people at/near Caltech,
there will be a special seminar at 2pm today in 248 Lauritsen.
Makoto Sakuda of KEK, Japan will present the preliminary results
from K2K long baseline neutrino Oscillation Experiment.
If I were there, I would definitely go. Anyone cou |
|
h******e
发帖数: 1791 | 28 我正在分析一个简单的dataset，用linear regression分析欧洲的食品消费问题，以肉
类作response，其他各类食品作为predictor。从数据上看东欧的肉类消费明显小于西
欧，如果以东欧作为baseline，西欧作为categorical variable的dummy variable带入
model后，西欧的parameter是负的；如果以西欧作为baseline，东欧以同样的方式带入
model后，东欧的parameter是正的。该如何解释这个model呢？谢谢。 |
|
g******c
发帖数: 116 | 29 １。 有３个时间点，BASELINE， TIME１， TIME２
２。 每个病人BASELINE有各类变量。
３。 TIME１，TIME２上有EVENT／结果（０／１）。也就是TIME１成功=１ 或TIME１失
败=０；TIME２成功=１ 或TIME２失败=０；
４。 TIME１成功了就不会有TIME２。
５。 然后TIME１这个时间段有变量X＿TIME１；这个X＿TIME１对TIME１的成功与否有
影响。
６。 TIME２这个时间段有变量Y＿TIME２；这个Y＿TIME２对TIME２的成功与否有影响。
不知道什么模型有能符合这个数据？我想过DISCRETE SURVIVAL，有两个时间点和基线
变量。但不知道如何处理X＿TIME１，Y＿TIME２变量。他们不是TIME－VARYING
COVARIATE因为他们只固定各自的时间段。而且他们对各自时间点的结果确有不可忽略
的影响 |
|
d******e
发帖数: 7844 | 30 当样本数量很大的时候，这么做没有问题，当样本少的时候就有麻烦了。尤其是高维数
据。
其实普通的KNN overfitting就是很严重，而我也只是拿KNN来做baseline。我导师非要
我弄一个overfitting最小的训练策略然后用这个数据来做baseline。 |
|
D******n
发帖数: 2836 | 31 i guess for model 2 u should dump all the variables to fit a new model, rest
ricting the coefficients is weird,
., |
|
j*****e
发帖数: 182 | 32 Try this:
proc logistic data=A;
model event = X3/offset=z;
run; |
|
d**********r
发帖数: 24123 | 33 举个例子吧：
在 time 0 的时候，Control 骨密度是 1， 在 time 3 month 的时候 Control 骨密度
是 2
在 time 0 的时候，Mutant 骨密度是 5， 在 time 3 month 的时候 Mutant 骨密度是
3
那么就用 Time 0 的数据作为Baseline，计算 Change of Baseline 为：
CFB = 2-1 = 1 ， control
CFB = 3-5 = -2 , Mutant
然后，把 CFB 作为response:
Mixed Model:
CFB = treatment + 其他factors
这样来看 treatment的effects |
|
b***8
发帖数: 60 | 34 In biomarker studies, the ultimate goal is to screen dozens, or
hundreds, or potentially hundreds of thousands of potential predictors
using explanatory methods to understand the existing data and then
ultimately identify a factor or set of factors that could be
used in the prediction of new observations with high accuracy. While
the concepts of "association testing" and "classification" are not
unrelated in this scenario, they are inherently separate objectives
requiring the implementation of d... 阅读全帖 |
|
e**t
发帖数: 1 | 35 想用data step建立一个新data, 需要两个新变量 baseline and year_followed
原data是这样的
School Year grade
1 2005 03
1 2005 04
1 2005 05
1 2006 03
1 2006 04
1 2006 05
想建立这样的data
School Baseline Year_followed Year grade
1 2005 2005 0 03
1 2005 2005 0 04
1 2005 2005 0 05
1 2006 2005 1 03
1 2006 2005 1 04
1 2006 2005 1 05
另外想问下年级(grade) 每年... 阅读全帖 |
|
Y******Y
发帖数: 8753 | 36 第一个model用age作为time scale, 它是通过对entry age做left truncation以及
assume baseline hazard is a function of age来adjust for age. 并且baseline
hazard里面对age effect没有做任何的parametric assumption. 所以在第一个model里
面你是不需要另外将age当作是covariate来adjust。
第二个model是用time on study作为time scale,这个你要adjust for age as a
covariate.这个model你是assume age的effects是exponentially。
两个model都是valid的。但是要注意这两个models大部分时候不是等价的，如果你耐心
写出他们的partial likelihood,他们是不一样。
hope this helps. |
|
s*****9
发帖数: 108 | 37 治疗前的那些数据就属于baseline啊。简单的做法就是比较两个group治疗前后的各项
数据的差，复杂些的就是做model的时候把baseline的这些当做predictor。 |
|
w******8
发帖数: 59 | 38 I am so confused about what to include in the ITT population:
1) Some patients were randomized and got baseline measurements but never got
treatment and then dropped out. I don't think they need to be included
because their dropping out have nothing to do with the treatment.
2) Some patients drop out specifically for reasons not related to treatment.
For example, a patient dropped out because he found out he has another
disease and would like to focus on treatment of that disease. And this
patie... 阅读全帖 |
|
d***2
发帖数: 341 | 39 ITT means the subject is to receive study treatment, therefore any
randomized subject is included in the ITT population.
There are times companies use Modified ITT (MITT) for efficacy analyses,
this population includes subjects who have data for the efficacy
analyses (
ex..have both baseline and post-baseline data).
ITT has nothing to do with drop-outs, nor the reason of drop-outs.
Subjects mentioned in your (1) and (2) would be excluded from the safety
population since they never received study... 阅读全帖 |
|
x**********0
发帖数: 2260 | 40 data set是病人数据，分成baseline，month1 和month12 3个部分。
随着时间推移，病人有drop的。所以数据越来越少。（最后剩80左右）
一个月的数据中对照组有些重复测量了一次。即两个测量值/病人。
介绍：
病人随机分成对照组和试验组两个大组。
有两个测量指标，mobility 和 memory.
测量分别在Baseline (prior to intervention), then at Month 1 and Month 12.
要求分析：
两组间在 Month 1的mobility结果上有没有区别? How about across both Month 1
and Month 12?
两组间在 Month 1的memory结果上有没有区别? How about across both Month 1 and
Month 12?
在mobility的improvement和memory的improvement之间有没有correlation? (Lower
scores are better.)
请教怎么用SAS解决这些问题。思路是什么样的。要做些什么图... 阅读全帖 |
|
k****i
发帖数: 347 | 41 为了5个包子，上了
-----------两组间在 Month 1的mobility结果上有没有区别?
用t-test (continuous outcome) 或 chi-square (categorical outcome) 就可以了
proc ttest;
proc freq;
-----------How about across both Month 1 and Month 12?
考虑到是repeated measure，用linear mixed model（对continuous outcome）
或者generalized linear mixed model（对dichotomous或categorical outcome）
proc mixed data=;
class month id;
model mobility = treatment / solution;
repeated month / subject=id type=un;
run;
------------两组间在 Month 1的memory结果上有没有区别? How about across... 阅读全帖 |
|
w******8
发帖数: 59 | 42 We are doing a clinical trial where there is significant missing data at
follow up. Some were even missing baselines. ITT is the proposed analysis.
We used LOCF to impute missing data but this would not help missing
baselines. Shall I use more fancy MI technique? I haven't touched MI before
though.
Also, since ITT is the proposed analysis, when I do a response profile (with
mean and SE over the time), shall I present the imputed data or the
available cases (not just completers)? Or both?
Thanks ... 阅读全帖 |
|
A*******s
发帖数: 3942 | 43 age cannot be used as a time-dependent effect since it will be absorbed in
the baseline hazard function.
but baseline age can be used as a time-independent effect. |
|
p*******i
发帖数: 1181 | 44 如果是两两时间比较（比如都比较Change from baseline) 楼上说的很正确， 或者如
果是小样本的话用 Fisher Exact test 也不错~
如果你是要overall的话 试着把后面每个时间点的dietary换成 1 if same as
baseline, 0 otherwise. 然后用Logistic Regression， 看看time是不是significant |
|
y*****w
发帖数: 1350 | 45 This could be a split-plot design. Treat DIET as the whole unit and TIME (
baseline, 6 mo, 12 mo) as the sub-units. Use the count (frequency) of each
diet within each id at each time point as the outcome variable. I’ve made
up a data as follows, assuming that at each time point an id could have any
combinations of diets but the total number of meals is always 4 (e.g. 2
meals of chicken, 0 meal of turkey, 2 meals of ham, 0 meal of fish for ID
#2 at baseline):
/*
coding of diet - chicken 1
... 阅读全帖 |
|
y*****w
发帖数: 1350 | 46 Thanks for all the comments and suggestions. It seems mixture modeling is a
very effective way to handle this type of situation. However, not only do I
need a model for that variable to be an outcome variable, but I also need to
have the variable controlled for as a covariate in another model. In the
latter scenario, I think I still need some way to transform the variable --
is there any good way to do that? One example of the latter scenario is that
I want to create a model for the change in th... 阅读全帖 |
|
w*******n
发帖数: 469 | 47 here the interaction is between the covariate and time. You may give the
treatment effect by times. Or you can divide your data into baseline and non
-baseline, then do the mixed model analysis. |
|
c********h
发帖数: 330 | 48 对survival不是很熟，但我猜应该跟linear model里有intercept的contrast是一样的。
你有个baseline的hazard h0 （不清楚survival是把这个intercept放在h0里还是你的
beta里有个截距项？看你的code像是放在h0里了），这个baseline应该是overall
effect, no matter treatment = 0 or 1. beta对应的应该是 treatment = 1 minus
the overall effect，所以要减去。
不知道说清楚没，打个比方，就像是
beta0 * x.bar + beta1 *(x - x.bar) = beta1 * x
你这个code里是用等式左侧做的calculation，但是你coding时pass进去的时候用的是
等式右侧。 |
|
c*********y
发帖数: 122 | 49 Not sure your purpose.
Firstly, there is seasonality and credit policy fine-tune/change. Not sure
you can keep a fixed baseline for a long period time.
Secondly, your marketing department might fool you by targeting more peoples
below your cut-off which might cause higher applications than usual but low
approval rate. So it is hard to set a baseline.
Just my 2 cents. |
|
