A*******s
发帖数: 3942 | 1 Subset Selection in Regression这本书讲了不少细节，不过我觉得内容有点老。网上
有电子书。
简单来说GLM的automatic selection都是基于likelihood的，一般来说是likelihood
ratio test with pre-specified significance level，当然也可以用AIC，BIC，
cross validation之类的。
leaps
forward |
|
n*****t
发帖数: 41 | 2 谢谢，文件如下。我只要读取最后一个数，就是那个likelihood的数值。
MERLIN 1.1.2 - (c) 2000-2007 Goncalo Abecasis
References for this version of Merlin:
Abecasis et al (2002) Nat Gen 30:97-101 [original citation]
Fingerlin et al (2004) AJHG 74:432-43 [case selection for
association studies]
Abecasis and Wigginton (2005) AJHG 77:754-67 [ld modeling, parametric
analyses]
Fingerlin et al (2006) Gen Epidemiol 30:384-96 [sex-specific maps]
Chen and Abecasis (2007) AJHG 81:913-26 [qtl assoc... 阅读全帖 |
|
c*******o
发帖数: 8869 | 3 用logistic regression 得到了predicted probability(for case=1), 结果在
predicted likelihood 很高 (0.8-1) 的区间 和很低的区间 （0-0.1), case
enrichment 很高， 而在中间区 （0.5左右）， case 的比例很底， 为什么predicted
likelihood 会有这种curvature的情况， 如何处理？
跪谢了。 |
|
k*******a
发帖数: 772 | 4 我的理解是把data分成k份，做个for loop
对于j份sample， 对于其他k-1份sample做regression得到coefficient estimate
然后用这个estimate算这个j份的partial likelihood, lik_j
然后把所有的 lik_j 加起来得到总的cross validation的likelihood |
|
l*g
发帖数: 46 | 5 非常谢谢ls回答我，j是什么？您是说我把10个从转了一圈的k－1个sample做出来的
model的partial likelihood加起来，这就是cross－validation的likelihood，然后呢
？怎么算prediction error？
还有，不是应该每一个fold内部先k－1组出来一个model，然后用剩下的一组validate
一下么？
抱歉我概念不太清，请教这个validate应该怎么做。。。非常感谢！! |
|
k*******a
发帖数: 772 | 6 我觉得cox model 不好算prediction error把，用likelihood作为指标就可以了
哪个parameter给的cross validation的likelihood好就用哪个
你是做ridge吗？还是。。。
validate |
|
l*g
发帖数: 46 | 7 抱歉刚才点错变成回信了，请忽略！
我就是做的proportional hazard model，那么剩下那一组testing sample可以
evaluate the prediction likelihood么？怎么得到predicted likelihood呢？我是
follow一篇paper里说每一个fold里的9／10组sample做一个model，然后用stepwise来
从k个predictors里面选best model，然后用剩下的1／10组sample来evaluate the
prediction log-lik. "For each k, 把所有predicted log－lik加总得到一个score，
score越小的就越好”，没明白什么意思。。。
谢谢！ |
|
A*******s
发帖数: 3942 | 8 这篇文章解释了给定misclassification rate的情况下AUC的分布
http://books.nips.cc/papers/files/nips16/NIPS2003_AA40.pdf
可以看到AUC's std会因为rare event而变大
我不知道有没有类似的讨论Bernoulli likelihood和AUC的paper，不过直觉上
Bernoulli likelihood其实就是error rate的一个continuous and differentiable
approximation, 也许关系会比较类似，这也和我记忆里，做weight adjustment或者调
整cost function一般有助于提高AUC的经验（不过经验很有限）相符。 |
|
w******e
发帖数: 142 | 9 没有直接的解析式你就要把marginal的likelihood+copula的likelihood放在一起MLE，
可以采用 onestep或者two step的方法，第一个方法你最后参数可能很多有可能数值解
不稳定optimizer会崩盘，如果要自己骗自己那就用基因算法来求“最优解”。第二个
求解就要方便很多，但是做inference很烦很烦，你要修整很多东西，当时就是给老板
做第二个的inference做得吐火。 |
|
D*********Y
发帖数: 3382 | 10 O，我仔细看了看wiki里面的明白了。
它只是build了event=1的partial likelihood function。
event=0只是用calculate event=1的likelihood。 |
|
w*******9
发帖数: 1433 | 11 你说的是Cox Model. censored data 在partial likelihood 里只起到计算at risk
set 的作用。partial likelihood也是从小白兔说的那个方法来的，就是profile out
the discrete baseline hazard function. |
|
w*******9
发帖数: 1433 | 12 坐等大牛来解释。
个人觉得直接拿rate做regression不妥，首先每个rate基于的obs个数不一样，所以每
个rate的
reliability (variance) 不一样，更重要的是你的obs的个数怎么来的肯定能起到决定
性作用。obs不是Independent variable!
如果你能自己编程的话，可以考虑以下方法：
1) Given ni, the rate pi follows G(n_i). (e.g. Beta regression for
fractional data)
2) Given pi, Yij follows binomial B(pi)
3) 1)and 2) determines P(Yij>0|n_i)=int_0^1 pG(p,f(n_i))dp
4) Write down the conditional likelihood of Yij given ni, i=1,2,.., j=1, 2,
..ni.
5) Maximize the likelihood wrt to parameters contained in G(). Ass... 阅读全帖 |
|
h*********n
发帖数: 278 | 13 谢谢回复。我感觉跟penalize by number of predictors没太大关系，因为data很大，
log-likelihood本来就很大哇，那个2k对AIC根本没啥影响，AIC的增减都是以百／千计
的，主要还是对log-likelihood 的影响。试过只include俩完全不可能相关的
predictor,还是有同样的现象。我感觉要么是data的问题，一方面很多zero,另一方面
有一些extremely large number, 用了这个predictor后，有助于ranking, 但是make
point prediction反而更难，所以model fit 更差， AIC increase. Does this make
sense? Actually I'm not so sure.
另一个可能是，我的model assumption不对，应该换一种model? Could this be the
reason? 头大。。。
AIC是一个fitting to the data和model complexity的trade-off。增加一个predictor... 阅读全帖 |
|
A*******s
发帖数: 3942 | 14 这篇文章解释了给定misclassification rate的情况下AUC的分布
http://books.nips.cc/papers/files/nips16/NIPS2003_AA40.pdf
可以看到AUC's std会因为rare event而变大
我不知道有没有类似的讨论Bernoulli likelihood和AUC的paper，不过直觉上
Bernoulli likelihood其实就是error rate的一个continuous and differentiable
approximation, 也许关系会比较类似，这也和我记忆里，做weight adjustment或者调
整cost function一般有助于提高AUC的经验相符。 |
|
v*******e
发帖数: 11604 | 15 maximum likelihood
找本统计书，看懂maximum likelihood，你就可以动手写程序算算看了。 |
|
P*****6
发帖数: 273 | 16 如果不研究理论推倒，这个难点不在理解，而在于推出那个长长的公式。
先是joint likelihood
然后是n多conditonal likelihood(具体什么名忘了）错一个就完了
我记得MCMC有一个优点
： 能够允许含有高维hidden variable， 其他的如EM含有四个hidden variables一般
人就干不动了 |
|
c*******e
发帖数: 150 | 17 想请教一下版上的各位大牛们，如果
Linear Regression中Noise Term是一个AR(1) process，通常都有什么成熟的算法做
MLE 或者其它方法 fit ？
具体的说，模型可以表示为 Y(t) = X(t) \dot \beta + E(t),
X(t) 和 \beta 都是 K-维的向量，其它都是标量。
t = 1, 2, 3, ..., T 是手头的 sample，
但是和经典的 Linear Regression 不同，E(t) 不是 i.i.d. 的高斯白噪音，可以假定
E(t) 服从一下 model:
E(t) = \rho * E(t-1) + \sigma * Z(t)
\rho 和 \sigma 是 unknown parameter，Z(t) 可以认为是高斯白噪音。
所以全部的 parameters 包括 向量 \beta 和标量 \sigma, \rho
最好还是 maximum-likelihood 的方法，这样我可以保留后面做 log-Likelihood
Ratio
Test 的可行性，以便于 做 model comparison/s... 阅读全帖 |
|
Q*****T
发帖数: 558 | 18 lz对统计的迷惑程度估计已经被各位笑掉大牙了。。。其实我有挺多东西不懂（虽然其
实日常工作中也基本用不到），但是仍然很想搞清楚的。。。
1，被你们拍砖说GLM的residual不一定是正态分布以后，我google了一下，学习了
residual的分布跟response variable的分布相关，譬如data是binary，那residual就
是binomial。那么问题来了，http://www.mun.ca/biology/dschneider/b7932/B7932Final10Dec2008.pdf 这篇文章第二页第六行，说model fit improvement是chi－square distribution （关于这点我也是一知半解，我课上跟老师做过nested model comparison，就是用两个model的－2log likelihood的差，再用degree of freedom的差，用chisquare statistics比较两个model是不是有显著不同），然后这篇文章还是第二页，第13行说到“The importance of normality ... 阅读全帖 |
|
c***z
发帖数: 6348 | 19 【 以下文字转载自 DataSciences 讨论区 】
发信人: chaoz (没钱也任性), 信区: DataSciences
标 题: ［请教］一个R问题
发信站: BBS 未名空间站 (Sat Apr 25 14:31:06 2015, 美东)
我最近做life time value，用的BTYD包
http://cran.r-project.org/web/packages/BTYD/vignettes/BTYD-walk
里面主要的方程是
pnbd.EstimateParameters 它会call optim，使用L-BFGS-B method
pnbd.LL 这是Pareto/NBD的likelihood function
我能够replicate原文章的结果，但是公司自己的数据有这么三个issue
1. L-BFGS-B method requires a finite value for fn 应该是likelihood function
overflow了，我对所有数据先取一次log，基本上解决了这个问题
2. 非常非常的慢，即使我使用十分之一的数据，把data... 阅读全帖 |
|
s******y
发帖数: 522 | 20 像microarray和RNA-seq, 会产生成千上万个t-test statistics或者likelihood ratio
test statics
我现在想要找一个非genetic的数据也能产生大量的likelihood ratio test statics
或者有现成的test statistics直接拿来用最好
查文献好像只有imaging data符合要求, 但是我对imaging data的处理毫无经验
希望板上大牛能指点一些方向 |
|
c*******e
发帖数: 150 | 21 【 以下文字转载自 Statistics 讨论区 】
发信人: cavaliere (Un Baiser S'il Vous Plaît), 信区: Statistics
标 题: Regression中噪音项是一个AR(1)，如何做MLE或者其它Fit？
发信站: BBS 未名空间站 (Mon Sep 15 22:02:42 2014, 美东)
想请教一下版上的各位大牛们，如果
Linear Regression中Noise Term是一个AR(1) process，通常都有什么成熟的算法做
MLE 或者其它方法 fit ？
具体的说，模型可以表示为 Y(t) = X(t) \dot \beta + E(t),
X(t) 和 \beta 都是 K-维的向量，其它都是标量。
t = 1, 2, 3, ..., T 是手头的 sample，
但是和经典的 Linear Regression 不同，E(t) 不是 i.i.d. 的高斯白噪音，可以假定
E(t) 服从一下 model:
E(t) = \rho * E(t-1) + \sigma * Z(t)
\rho 和 \s... 阅读全帖 |
|
c*******e
发帖数: 150 | 22 Awesome. Upon doing further survey on this topic, I also think this is the
best solution.
Out of curiosity, may I ask a further questions:
given the sample X(t) and Y(t), suppose that beta_star(rho_star) maximized
the likehihood function of all beta given that rho == rho_star, and
rho_star(beta_star) maximized the likelihood function of all rho given that
beta == beta_star, namely this pair of beta_star(rho_star) and
rho_star(beta_star) is the fixed-point which we converged at step (3), is
there... 阅读全帖 |
|
e*******n
发帖数: 872 | 23 原题见
http://www.mitbbs.com/article_t/DataSciences/10029.html
专门开一个贴，尝试逐题解答。本人菜鸟，求大牛指导
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
1. Given a coin you don’t know it’s fair or unfair. Throw it 6 times and
get 1 tail and 5 head. Determine whether it’s fair or not. What’s your
confidence value?
我的答案是：
H0: the coin is fair
Ha: the coin is unfair
X is the number of heads
Rejection region: |X - 3| > 2, i.e., X = 0,1,5,or 6
significance level alpha:
alpha = P(reject H0 | H0 i... 阅读全帖 |
|
c****t
发帖数: 19049 | 24 gelman的bda是大号经典没错但是出了名的难读难懂。等你从别的地方全学会了回头拿
这当reference可以。没基础的话看看jeff gill的bayesian methods吧
其实找工作刷题才是王道(不只是leetcode)，进工业界干活不用懂这么多。多讨论面试
题更有用。统计版不让讨论面试题也是我开此版的主要原因之一。
另外mcmc是从理论物理来的。跟统计一根毛的关系都没有。真想学懂直接看yann lecun
有关energy based models的东东也许能明白的更快一些。我一向认为灌统计水的大谈
特谈bayesian,扯神马frequenists vs. bayesian probability，神马prior,
postorior,就是为了掩盖mcmc根本不是统计创造发明出来的事实。一般人从神马prior,
posteriror啊，bayesian probability啊学起，等你学到mcmc时早迷糊了。也就不会
多问了
mcmc是跟传统的maximum likelihood完全不同的体系。maximum likelihood说简单点就
是做优化。优化的思路就是求... 阅读全帖 |
|
d******e
发帖数: 7844 | 25 他讲的就是Maximum Likelihood和Bayesian Estimator的基本思想和基本区别。
一个是在是求最大化参数Likelihood。
至于说用Gradient Ascent，用Newton。Stochastic做，Batch做不过都是手段罢了。
一个是对Posterior做积分。
至于如何Sampling，是Gibbs Sampling, 还是更general的MH之类的，也只是手段罢了。
呵呵，你这是要Challenge我么？ |
|
c****t
发帖数: 19049 | 26 gelman的bda是大号经典没错但是出了名的难读难懂。等你从别的地方全学会了回头拿
这当reference可以。没基础的话看看jeff gill的bayesian methods吧
其实找工作刷题才是王道(不只是leetcode)，进工业界干活不用懂这么多。多讨论面试
题更有用。统计版不让讨论面试题也是我开此版的主要原因之一。
另外mcmc是从理论物理来的。跟统计一根毛的关系都没有。真想学懂直接看yann lecun
有关energy based models的东东也许能明白的更快一些。我一向认为灌统计水的大谈
特谈bayesian,扯神马frequenists vs. bayesian probability，神马prior,
postorior,就是为了掩盖mcmc根本不是统计创造发明出来的事实。一般人从神马prior,
posteriror啊，bayesian probability啊学起，等你学到mcmc时早迷糊了。也就不会
多问了
mcmc是跟传统的maximum likelihood完全不同的体系。maximum likelihood说简单点就
是做优化。优化的思路就是求... 阅读全帖 |
|
d******e
发帖数: 7844 | 27 他讲的就是Maximum Likelihood和Bayesian Estimator的基本思想和基本区别。
一个是在是求最大化参数Likelihood。
至于说用Gradient Ascent，用Newton。Stochastic做，Batch做不过都是手段罢了。
一个是对Posterior做积分。
至于如何Sampling，是Gibbs Sampling, 还是更general的MH之类的，也只是手段罢了。
呵呵，你这是要Challenge我么？ |
|
c***z
发帖数: 6348 | 28 我最近做life time value，用的BTYD包
http://cran.r-project.org/web/packages/BTYD/vignettes/BTYD-walk
里面主要的方程是
pnbd.EstimateParameters 它会call optim，使用L-BFGS-B method
pnbd.LL 这是Pareto/NBD的likelihood function
我能够replicate原文章的结果，但是公司自己的数据有这么三个issue
1. L-BFGS-B method requires a finite value for fn 应该是likelihood function
overflow了，我对所有数据先取一次log，基本上解决了这个问题
2. 非常非常的慢，即使我使用十分之一的数据，把data frame变成matrix也没有帮助
3. 结果不stable，几乎总是取得local optimal，我试着用了 mothod = SANN (
simulated annealing) 也没有帮助
有点stuck了，所以想问问大家
多谢多谢！
cha... 阅读全帖 |
|
f****X
发帖数: 64 | 29 $CLSN
The HEAT study interim analysis results, likely due 8 weeks after the 190+
PFS announcement. (Maybe, by end of October, early November at the very
latest).
This is obviously the catalyst many investors are thinking about, and
rightfully so. At this point, with a chance for futility being "de minimis"
to quote the CEO, I think we can safely assume we are looking at either a
continuation to 380 events, or overwhelming efficacy sufficient to justify
an early NDA. While the study was designed... 阅读全帖 |
|
q********g
发帖数: 10694 | 30 观点很象这里Jenny12以前说的。从中也学习到一些分析医药股的方法。例文欣赏。
http://seekingalpha.com/article/895051-sarepta-therapeutics-loo
Sarepta Therapeutics Looks Highly Undervalued
September 28, 2012 | 37 commentsby: Sean Anderson | about: SRPT
R.S. Analytics wrote up the long case for Sarepta Therapeutics (SRPT)
earlier this month here. I agree with that thesis, and won't bore everyone
by repeating it here since R.S. did such a great job telling it already.
There are, however, a few points that I'd like to expand on.
In ... 阅读全帖 |
|
w******s
发帖数: 16209 | 31 The United States Patent and Trademark Office has issued a refusal on Google
's Nexus One application.
According to nexus-one.co.uk Google filed an application for the trademark
back on 10th December 2009 for use in connection with mobile phones.
The Trademark Office recently issued a notice of refusal:
SECTION 2(d) REFUSAL ? LIKELIHOOD OF CONFUSION
Registration of the applied-for mark is refused because of a likelihood of
confusion with the mark in U.S. Registration No. 3554195.
"The refusal ha |
|
w******s
发帖数: 16209 | 32 The United States Patent and Trademark Office has issued a refusal on Google
's Nexus One application.
According to nexus-one.co.uk Google filed an application for the trademark
back on 10th December 2009 for use in connection with mobile phones.
The Trademark Office recently issued a notice of refusal:
SECTION 2(d) REFUSAL ? LIKELIHOOD OF CONFUSION
Registration of the applied-for mark is refused because of a likelihood of
confusion with the mark in U.S. Registration No. 3554195.
"The refusal ha... 阅读全帖 |
|
j***b
发帖数: 5901 | 33 Didn't I say the cctv video is evidence?
If the woman presented receipt for the ring, it should be considered
reasonable evidence. The judge may also have asked the guy to describe the ring, and the description matched that of the expensive ring.
It could still be con job, but the likelihood is low. That should be good
enough for such civil trial. |
|
w*******q
发帖数: 1764 | 34 well, if you call it system, it needs to be written into laws to formalize
its position and legitimacy, otherwise, there is the likelihood that any
time any one of the nth generation leader feels not entirely satisfied with
his successor, something might happen. Sure, this time, nothing serious
happens, but I think everybody remembers the chatter about Hu being shackled
by Jiang back in early 2000s. The fact of matter is there needs to be some
sort of check and balance, transparent politics in p |
|
k*****r
发帖数: 21039 | 35 两个理工科男，其实是先有了结论和观点，再找对自己有利的数据。
这个和俺的先有事实和数据，再形成max likelihood hypothesis的apporach完全是南
媛北扯。 |
|
D*****i
发帖数: 8922 | 36 再来点干货。《科学》发表的，直接测仰韶文化的距今天5700年的古绵羊骨头。这些绵
羊的食物中一部分是小米。作者认为可以合理地说这些是驯化的绵羊。
http://www.nature.com/articles/srep07170
The identification of the bones on physical characteristics and DNA fragment
confirms the bone samples are from the genus Ovis (sheep). The radiocarbon
age of 5700 BP (4700 to 4400 BCE) for the three bones from Shihushan is the
oldest independent age for sheep in eastern Asia. Since the bones were found
in association with other domestic species and in an archaeological setting
of th... 阅读全帖 |
|
D*****i
发帖数: 8922 | 37 再来点干货。《科学》发表的，直接测仰韶文化的距今天5700年的古绵羊骨头。这些绵
羊的食物中一部分是小米。作者认为可以合理地说这些是驯化的绵羊。
http://www.nature.com/articles/srep07170
The identification of the bones on physical characteristics and DNA fragment
confirms the bone samples are from the genus Ovis (sheep). The radiocarbon
age of 5700 BP (4700 to 4400 BCE) for the three bones from Shihushan is the
oldest independent age for sheep in eastern Asia. Since the bones were found
in association with other domestic species and in an archaeological setting
of th... 阅读全帖 |
|
发帖数: 1 | 38 The Liberty Bell
Liberty Bell
Tradition tells of a chime that changed the world on July 8, 1776, with the
Liberty Bell ringing out from the tower of Independence Hall summoning the
citizens of Philadelphia to hear the first public reading of the Declaration
of Independence by Colonel John Nixon.
The Pennsylvania Assembly ordered the Bell in 1751 to commemorate the 50-
year anniversary of William Penn's 1701 Charter of Privileges, Pennsylvania'
s original Constitution. It speaks of the rights and... 阅读全帖 |
|
g**1
发帖数: 10330 | 39 http://www.cnbc.com//id/39567592
Moody's Puts China in Line for Possible Credit Upgrade
Published: Friday, 8 Oct 2010 | 12:36 AM ET Text Size
By: Reuters
Moody's Investors Service rewarded China on Friday for its response to
the global financial crisis by placing the government's bond rating on
review for a possible upgrade.
China's resilient economic performance, the prospect of continued strong
growth over the medium term and the likelihood that Beijing can handle
any losses from an unpreceden... 阅读全帖 |
|
z****n
发帖数: 5870 | 40 http://www.cnn.com/2010/BUSINESS/11/04/dalai.lama.trade.china/index.htm
l
China trade: The 'Dalai Lama Effect'
Beijing, China (CNN) -- Countries whose top leadership meet with the
Dalai Lama, Tibet's exiled spiritual leader, lose on average 8.1
percent in exports to China in the two years following the meeting,
according to a recent study.
Called the "Dalai Lama Effect," the study by the University of
Gottingen in Germany found the negative impact on exports began when
President Hu Jintao took o... 阅读全帖 |
|
R****a
发帖数: 6858 | 41 帝国的陨落
来源: ognc 于 2010-12-12 05:58:18 [档案] [博客] 旧帖] [转至博客] [给我悄悄话
] 本文已被阅读：254次 字体：调大/调小/重置 | 加入书签 | 打印 | 所有跟帖 | 加
跟贴 | 查看当前最热讨论主题 0
The Decline and Fall of the American Empire
By Alfred W. McCoy
December 11, 2010 "The Nation" - - A soft landing for America 40 years from
now? Don’t bet on it. The demise of the United States as the global
superpower could come far more quickly than anyone imagines. If Washington
is dreaming of 2040 or 2050 as the end of the American Century, a more
realistic asses... 阅读全帖 |
|
s**********1
发帖数: 305 | 42 A strong evidence is that the genes of homo sapien sapiens, i.e., us,
outside of Africa are only similar to a subpopulation of those in Africa.
This is a very strong support that we are all originally from Africa, not
vice versa. The likelihood that human race evolved first outside of Africa
and maintained relative genetic homogeneity during the evolution, then went
to Africa and gained tons of mutations is minimal. Genetics studies are hard
evidence, other evidence include linguistics and cu... 阅读全帖 |
|
M*****8
发帖数: 17722 | 43 http://bit.ly/hV0fzM
注意读者的评论。
Indian-American gets 32 yrs for selling US secrets
AGENCIES, Jan 26, 2011, 07.22am IST
WASHINGTON: Indian-American Noshir Gowadia, a former B-2 stealth bomber
engineer, was on Monday sentenced by a US court to 32 years in jail for
selling American military secrets to China.
Gowadia, 66, has spent the past five years in prison. Prosecutors alleged
that the Mumbai-born Gowadia, who showed no emotions as the judge announced
the punishment, had helped design an exhaust ... 阅读全帖 |
|
p******u
发帖数: 14642 | 44 看这洋鬼子多悲愤，好像我鳖又偷了他家的鸡还啥的
http://nucleargreen.blogspot.com/2011/01/have-chinese-been-read
Have the Chinese Been Reading Energy from Thorium or Nuclear Green?
Last week the Chinese Academy of Science announced that it planned to
finance the development of a Chinese Thorium Breeding Molten Salt Reactor (
TMSR) or as it is called in the United States, the Liquid Fluoride Thorium
Reactor (LFTR). The announcement came in a news report from Weihui.news365.
com.cn. The announcement was relayed to Western... 阅读全帖 |
|
j***h
发帖数: 4412 | 45 Earthquake detection systems can sound the alarm in the moments before a
big tremor strikes—time enough to save lives
Japan’s system, which went live in 2007, makes heavy use of personal
technology. Alerts go out not only on television and radio but through
special receivers in homes, offices and schools. Pop-up windows on
computers show a real-time map with the epicenter’s location and the
radiating seismic waves. A timer counts down to the shaking at your
location and highlights predicted inte... 阅读全帖 |
|
c******n
发帖数: 5697 | 46 【 以下文字转载自 WaterWorld 讨论区 】
发信人: chairman ( 8====o ~~ /*\), 信区: WaterWorld
标 题: 核灾难已不可避免：日本放弃抢救核电站已全面撤退
发信站: BBS 未名空间站 (Mon Mar 14 22:25:22 2011, 美东)
日本首相11am东京时间要发布讲话了
TOKYO — Japan faced the likelihood of a catastrophic nuclear accident
Tuesday morning, as an explosion at the most crippled of three reactors at
the Fukushima Daichi Nuclear Power Station damaged its crucial steel
containment structure, emergency workers were withdrawn from the plant,
and much larger emissions of radioactive ma... 阅读全帖 |
|
d****n
发帖数: 10034 | 47 【 以下文字转载自 WaterWorld 讨论区 】
发信人: chairman ( 8====o ~~ /*\), 信区: WaterWorld
标 题: 核灾难已不可避免：日本放弃抢救核电站已全面撤退
发信站: BBS 未名空间站 (Mon Mar 14 22:25:22 2011, 美东)
日本首相11am东京时间要发布讲话了
TOKYO — Japan faced the likelihood of a catastrophic nuclear accident
Tuesday morning, as an explosion at the most crippled of three reactors at
the Fukushima Daichi Nuclear Power Station damaged its crucial steel
containment structure, emergency workers were withdrawn from the plant,
and much larger emissions of radioactive ma... 阅读全帖 |
|
u***r
发帖数: 4825 | 48 http://www.nytimes.com/2011/05/16/nyregion/maid-picks-imf-chief
By MOSI SECRET
Published: May 15, 2011
* Recommend
* Twitter
* Sign In to E-Mail
* Print
*
Reprints
* ShareClose
o Linkedin
o Digg
o MySpace
o Permalink
o
Dominique Strauss-Kahn, the leader of the International Monetary Fund, spent
most of Sunday at the Manhattan Special Victims Unit in East Harlem as
prosecutors sought additional evidence, including possible DNA ... 阅读全帖 |
|
d**o
发帖数: 864 | 49 有时候就做一些简单的regression啊.maximum likelihood estimation.
调用超简单. |
|
b*******y
发帖数: 4304 | 50 if you want to make such a contract with part supplier such that they will
shoulder your business loss if their parts fail, you will have a very
expensive contract
by economy law, your business will naturally fail since your business will
have very high cost than those business simply using insurance.
This is, basically, a pool-sharing-loss(or insurance) question, just with
another face. The object has very low likelihood to fail, but do fail
sometimes. |
|
