Z*******0
发帖数: 11 | 1 是否考虑East Brunswick？这里有直接到NYC Port Authority的bus,大约45分钟（直接
上NJ Turnpike）。学区很好，华人比例较高。
推荐你查阅https://www.neighborhoodscout.com/nj/east-brunswick/
East Brunswick is a medium-sized town located in the state of New Jersey.
With a population of 46,125 people and ten constituent neighborhoods.
East Brunswick is a decidedly white-collar town, with fully 88.61% of the
workforce employed in white-collar jobs, well above the national average.
Overall, East Brunswick is a town of professionals, sales and off... 阅读全帖 |
|
Z*******0
发帖数: 11 | 2 是否考虑East Brunswick？这里有直接到NYC Port Authority的bus,大约45分钟（直接
上NJ Turnpike）。学区很好，华人比例较高。
推荐你查阅https://www.neighborhoodscout.com/nj/east-brunswick/
East Brunswick is a medium-sized town located in the state of New Jersey.
With a population of 46,125 people and ten constituent neighborhoods.
East Brunswick is a decidedly white-collar town, with fully 88.61% of the
workforce employed in white-collar jobs, well above the national average.
Overall, East Brunswick is a town of professionals, sales and off... 阅读全帖 |
|
l*********o
发帖数: 10 | 3 今天早上组里开会，有消息放出新工作机会，请大家务必抓紧申请！请电邮最近简历至
s************[email protected]
请注明title身份(opt,green card,或者h1)
JOB TITLE: SAS programmer (Openings: 1)
LOCATION: Bridgewater, NJ
TYPLE: Full-time
SALARY RANGE: 75,000-80,000
Must-haves:
Should be familiar with Graph, Macro, Base, analysis procedures,
Programmer with SAS versions 8.2
MUST BE ABLE TO CUSTOM CODE
Producing Data files
Manipulating statistical procedures
Creating graphs
Providing analysis for clinical trials and tabulating data/ writing in
medical j... 阅读全帖 |
|
s*******e
发帖数: 4188 | 4 又到了人口普查的时间了，其实这个人口普查跟IT巨头IBM还有一段渊源，不知大家听说过没有？
美国的人口普查10年一次。1880年，美国人口达到5千万。由于那时人口普查数据还需要人工统计，所有数据花了七年时间才统计完——这都快到了下次人口普查的时间了。由于人口的增长和统计内容的增加，人们普遍认为1890年的人口普查要花13年的时间才能统计完——这样迟到的数据已经没什么用了。
于是美国人口普查局找到了Herman Hollerith。Hollerith是学工程出身，他受到当时火车检票员的启发，发明了一种读卡机。当时的火车检票员会根据乘客的起始车站，性别，年龄，头发颜色等信息，在车票的不同位置打孔。Hollerith把这种机制做成了自动的机器。
1890年的人口普查就使用了他发明的读卡机，用了两年半就统计完了。其中总人口数只用了六个星期就得到了结果，让所有人都非常吃惊，很多人甚至认为他的结果是错误的。
Hollerith于1896年创办了自己的公司，Tabulating Machine Company，专门为各国人口普查和保险公司等造读卡机。这家公司于1911年和另外几家公司合并，成立了Co |
|
a*****3
发帖数: 10373 | 5 赞技术贴!
听说过没有？
需要人工统计，所有数据花了七年时间才统计完——这都快到了下次人口普查的时间了
。由于人口的增长和统计内容的增加，人们普遍认为1890年的人口普查要花13年的时间
才能统计完——这样迟到的数据已经没什么用了。
时火车检票员的启发，发明了一种读卡机。当时的火车检票员会根据乘客的起始车站，
性别，年龄，头发颜色等信息，在车票的不同位置打孔。Hollerith把这种机制做成了
自动的机器。
只用了六个星期就得到了结果，让所有人都非常吃惊，很多人甚至认为他的结果是错误
的。
人口普查和保险公司等造读卡机。这家公司于1911年和另外几家公司合并，成立了
Computing Tabulating Recording Corporation（CTR），CTR则于1924年改名为IBM。 |
|
S*********g
发帖数: 24893 | 6 http://www.prisoners.com/relcrime.html
宗教对犯罪率的影响
更加信仰基督的地方有更多的社会弊病和犯罪。这是格雷戈里。保罗在学术杂志《宗教
与社会》上发表的初步结论，这个杂志是内布拉斯加州一所教会学校的学报。
http://moses.creighton.edu/JRS/2005/2005-11.html
研究表明，基督教统治的美国南部和中西部地区，社会弊端和犯罪更严重，谋杀，性病
，死亡率更高。这项研究还表明，美国的犯罪率和基督徒比例，比其他发达世界国家都
高得多。
无法回避的现实是，基督教的教条更合适一个病态的社会。宗教狂热分子和极端主义比
毒品危害更大。
理性的人认识到基督教神话的荒谬可怜，而基督徒，尤其是福音派，或原教旨主义者，
不过是一些伪君子，用宗教作为武器来贬低他人。
格雷戈里。保罗的研究首次量化了基督教教义产生的社会效果。不只是基督教，一切狂
热的邪教都可能会有类似的邪恶效果。
当然，天主教会花了几百年压迫陷入贫困和苦难的老百姓。穆斯林有同样的呼吁
：暴力，复仇，排斥和压迫。
理智的人会奇怪宗教神话的诞生。
在美国，宗教=残酷。
Cr... 阅读全帖 |
|
s********n
发帖数: 1540 | 7 德州国买车也可以抵税，你说呢？
The American Jobs Creation Act of 2004 authorized the sales tax deduction as
an option for those who itemize deductions, letting them choose between
deductions for state and local income taxes or sales taxes. Taxpayers will
indicate by a checkbox on line 5 of Schedule A which type of tax they’re
claiming.
The tables give taxpayers a sales tax deduction amount as an alternative to
saving their receipts throughout the year and tabulating the amount actually
paid. Taxpayers use thei... 阅读全帖 |
|
M*******c
发帖数: 4371 | 8 【 以下文字转载自 USANews 讨论区 】
发信人: lczlcz (lcz), 信区: USANews
标 题: 美国新的特权阶级: 非法移民
发信站: BBS 未名空间站 (Mon Jan 12 17:50:08 2015, 美东)
The New Privileged Class: Illegal Immigrants
By Michael Bargo Jr.
The belief is widely held that those who are wealthy or possess great
political power are above the law and do not have to respond to the legal
constraints endured by the great masses of Americans. Those on the left
who profess to support the poor and middle classes are quick to point out
that the wealthy cla... 阅读全帖 |
|
l*********o
发帖数: 10 | 9 今天早上组里开会，有消息放出新工作机会，请大家务必抓紧申请！请电邮最近简历至
s************[email protected]
请注明title身份(opt,green card,或者h1)
JOB TITLE: SAS programmer (Openings: 1)
LOCATION: Bridgewater, NJ
TYPLE: Full-time
SALARY RANGE: 75,000-80,000
Must-haves:
Should be familiar with Graph, Macro, Base, analysis procedures,
Programmer with SAS versions 8.2
MUST BE ABLE TO CUSTOM CODE
Producing Data files
Manipulating statistical procedures
Creating graphs
Providing analysis for clinical trials and tabulating data/ writing in
medical j... 阅读全帖 |
|
B*****e
发帖数: 9375 | 10
没mvp是硬伤么?
我们不能看着大曼宁手里一堆的MVP，
将来肯定也是第一票入选，就把两者强烈联系起来。
这些二战后入堂的两毛五， 很多MVP? I did not tabulate.
Modern Era: Quarterbacks (23)
Troy Aikman 1989-2000
George Blanda (Also PK) 1949-1958, 1960-1975
Terry Bradshaw 1970-1983
Len Dawson 1957-1975
John Elway 1983-1998
Dan Fouts 1973-1987
Otto Graham 1946-1955
Bob Griese 1967-1980
Sonny Jurgensen 1957-1974
Jim Kelly 1986-1996
Bobby Layne 1948-1962
Dan Marino 1983-1999
Joe Montana 1979-1994
Warren Moon 1984-2000
Joe Namath 1965-1977
Bart Starr 1956-1971
Rog... 阅读全帖 |
|
s*****h
发帖数: 44903 | 11 ☆─────────────────────────────────────☆
siriusliu (天狼) 于 (Wed Oct 9 01:51:59 2013, 美东) 提到:
跟lp普及，聊起QB物以类聚，人以群分。跟大家分享，同求指教。
第一群，也是第一档，四大HOFer，菜鼻猪龙。ring也有，常规赛mvp也有（猪是例外，
但是N次前三，多次第二，而且是差不多可以并列第一的第二，算成mvp也没有问题）。
总之，无需再证明什么，HOF没跑。剩下的就是如何reinforce legacy。
第二群，小白菜，大本，伊力特。有运气也有心脏。有ring。但是从未常规赛打出mvp
水准。问题，这三人有无可能混进hof？多半没戏，但是有无可能？
第三群，麻软，小河，肉末，微克，杰小卡，四大夫，纱布。常规赛有过出色数据跟战
绩，也混了大合同，也算是franchise qb。但是从未季后赛证明自己。也老大不小了，
基本不可能撞上狗屎运挤入第二群。
第四群，四大新人，运气，rw，ck，rg3。勉强可以加进牛顿。还年轻，一起都早。
luck有望进第一群，rw跟ck有望进第二群。rg3跟... 阅读全帖 |
|
k*****9
发帖数: 516 | 12 这个时代周刊的搞笑节目。
把宇宙学校八丐排如此低？
http://time.com/4147924/college-football-top-25-ranked-by-acade
Northwestern celebrates winning the "Land of Lincoln" trophy after the
Wildcats defeated Illinois on November 28, 2015 in Chicago, Illinois.
And the winner is ...
Northwestern has plenty to celebrate. The Wildcats football team, far from a
traditional powerhouse, finished the regular season with a 10-2 record, and
earned a trip to the Outback Bowl on New Year’s Day.
And though Northwestern fell short of a n... 阅读全帖 |
|
n********r
发帖数: 2228 | 13 This is what I am looking for. Thanks.
relative risks, associated with
competitive sports.
event, and controversy persists regarding
systematically tabulated groups of
and Twin Cities (1982 to 1994) |
|
a***a
发帖数: 818 | 14 10 American Foods that are Banned in Other Countries
ARTICLE TAGS
NutritionHealth
AuthorDr. Mercola
Americans are slowly waking up to the sad fact that much of the food sold in
the US is far inferior to the same foods sold in other nations. In fact,
many of the foods you eat are BANNED in other countries.
Here, I’ll review 10 American foods that are banned elsewhere.
Seeing how the overall health of Americans is so much lower than other
industrialized countries, you can’t help but wonder whether... 阅读全帖 |
|
b*s
发帖数: 82482 | 15 刚才在路上，说不明白。除了那本书以外，原来在纽约客上Galdwell的文章可以给出详
细一些的例子吧：
http://www.gladwell.com/2008/2008_10_20_a_latebloomers.html
Articles from the New Yorker
Late Bloomers
October 20, 2008Annals of Culture
Why do we equate genius with precocity?
1.
Ben Fountain was an associate in the real-estate practice at the Dallas
offices of Akin, Gump, Strauss, Hauer & Feld, just a few years out of law
school, when he decided he wanted to write fiction. The only thing Fountain
had ever published was a law-review article. H... 阅读全帖 |
|
c*****s
发帖数: 180 | 16 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
c*****s
发帖数: 180 | 17 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
c*****s
发帖数: 180 | 18 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
c*****s
发帖数: 180 | 19 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
c*****s
发帖数: 180 | 20 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
c*****s
发帖数: 180 | 21 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
c*****s
发帖数: 180 | 22 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
c*****s
发帖数: 180 | 23 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
c*****s
发帖数: 180 | 24 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
c*****s
发帖数: 180 | 25 SAS OnlineTutor®: Advanced SAS®
Controlling Memory Usage 15 of 21
backnextlesson menuLearning Pathhelp menu
Using the SASFILE Statement (continued)
Bar chart iconComparative Example: Using the SASFILE Statement
Suppose you want to create multiple reports from SAS data files that vary in
size. Using small, medium, and large data files, you can compare the
resource usage when the PRINT, TABULATE, MEANS, and FREQ procedures are used
with and without the SASFI |
|
S*********g
发帖数: 24893 | 26 http://www.prisoners.com/relcrime.html
宗教对犯罪率的影响
更加信仰基督的地方有更多的社会弊病和犯罪。这是格雷戈里。保罗在学术杂志《宗教
与社会》上发表的初步结论，这个杂志是内布拉斯加州一所教会学校的学报。
http://moses.creighton.edu/JRS/2005/2005-11.html
研究表明，基督教统治的美国南部和中西部地区，社会弊端和犯罪更严重，谋杀，性病
，死亡率更高。这项研究还表明，美国的犯罪率和基督徒比例，比其他发达世界国家都
高得多。
无法回避的现实是，基督教的教条更合适一个病态的社会。宗教狂热分子和极端主义比
毒品危害更大。
理性的人认识到基督教神话的荒谬可怜，而基督徒，尤其是福音派，或原教旨主义者，
不过是一些伪君子，用宗教作为武器来贬低他人。
格雷戈里。保罗的研究首次量化了基督教教义产生的社会效果。不只是基督教，一切狂
热的邪教都可能会有类似的邪恶效果。
当然，天主教会花了几百年压迫陷入贫困和苦难的老百姓。穆斯林有同样的呼吁
：暴力，复仇，排斥和压迫。
理智的人会奇怪宗教神话的诞生。
在美国，宗教=残酷。
Cr... 阅读全帖 |
|
R******d
发帖数: 5739 | 27 http://movies.yahoo.com/blogs/movie-news/wolf-wall-street-drops
‘Wolf of Wall Street’ Drops Most Movie F-Bombs… Ever
So, how many times does "Wolf" drop the F-word? According to Wikipedia
tabulation: 506 times, or 2.83 times every 60 seconds of the 180-minute film
. This easily beat the previous record-holder among scripted films, Spike
Lee's 1999 hit "Summer of Sam," which boasted (a measly) 435 instances of
the expletive. |
|
s***s
发帖数: 7178 | 28 问题是我的帖子你看不见啊
About: Ramesses IIIAn Entity of Type : people, from Named Graph : http://dbpedia.org, within Data Space : dbpedia.org Ramesse III (... – ... ) è stato il secondo sovrano della XX dinastia egizia. |-bgcolor="EFEFEF" |align="center"|Nome Horo |align="center"|Altri nomi |- |align="center"|Ka-nekhet Aa-nesyt (numerose varianti) |align="center"|Ramesse III |}
Property Value
dbpedia-owl:abstract ■Ramses III. (* um 1221 v. Chr. ; † 7. April
1156 v. Chr. ) war ein altägyptische... 阅读全帖 |
|
S*********g
发帖数: 24893 | 29 http://www.prisoners.com/relcrime.html
宗教对犯罪率的影响
更加信仰基督的地方有更多的社会弊病和犯罪。这是格雷戈里。保罗在学术杂志《宗教
与社会》上发表的初步结论，这个杂志是内布拉斯加州一所教会学校的学报。
http://moses.creighton.edu/JRS/2005/2005-11.html
研究表明，基督教统治的美国南部和中西部地区，社会弊端和犯罪更严重，谋杀，性病
，死亡率更高。这项研究还表明，美国的犯罪率和基督徒比例，比其他发达世界国家都
高得多。
无法回避的现实是，基督教的教条更合适一个病态的社会。宗教狂热分子和极端主义比
毒品危害更大。
理性的人认识到基督教神话的荒谬可怜，而基督徒，尤其是福音派，或原教旨主义者，
不过是一些伪君子，用宗教作为武器来贬低他人。
格雷戈里。保罗的研究首次量化了基督教教义产生的社会效果。不只是基督教，一切狂
热的邪教都可能会有类似的邪恶效果。
当然，天主教会花了几百年压迫陷入贫困和苦难的老百姓。穆斯林有同样的呼吁
：暴力，复仇，排斥和压迫。
理智的人会奇怪宗教神话的诞生。
在美国，宗教=残酷。
Cr... 阅读全帖 |
|
y**e
发帖数: 2729 | 30 请看US 2011 10 best jobs 榜单
如果你是前3甲， 就吃包子
3甲只有1位英雄，第2位就不吃了
所以总计3歌包子，这年头刺激经济多难呀，政府也穷得叮当响呀
1. Software Engineer
Researches, designs, develops and maintains software systems along with
hardware development for medical, scientific, and industrial purposes.
Overall Score: 60.00Income: $87,140.00
Work Environment:
150.000
Stress:
10.400
Physical Demands:
5.00
Hiring Outlook:
27.40
2. Mathematician
Applies mathematical theories and formulas to teach or solve problems in a
business, educational, or indust... 阅读全帖 |
|
I****J
发帖数: 516 | 31 1. Software Engineer
Researches, designs, develops and maintains software systems along with
hardware development for medical, scientific, and industrial purposes.
Overall Score: 60.00Income: $87,140.00
Work Environment:
150.000
Stress:
10.400
Physical Demands:
5.00
Hiring Outlook:
27.40
2. Mathematician
Applies mathematical theories and formulas to teach or solve problems in a
business, educational, or industrial climate.
Overall Score: 73.00Income: $94,178.00
Work Environment:
89.720
Stress:
1... 阅读全帖 |
|
k****h
发帖数: 27 | 32 I have some trouble to output the text by ASP.
For example: I want to output a text table:
OrerId OrderItems Price
123 2 150
How do I make tab work? I only want it be text, not html. |
|
m**c
发帖数: 2103 | 33 这比那个是严谨可信一些。不过本着学术钻牛角尖的精神，该报告还有几点问题。仅供
娱乐哦。
首先这个第三方保险是自愿的，那就要考虑adverse selection.如果买该保险的用户无
法代表整个使用laptop的用户群那该报告就有问题。
其次该报告没有讲detail. 比如那个malfunction faliure rate, 难道就是把2年来的
统计数据tabulate一下？ 考虑到censoring至少估计下survival function吧，就算是
基本的parametric estimate with covariates.
这家公司很有意思，所有品牌的电脑保费都一样，只看价格。我瞅了一眼比联想自己的
质保贵不少，比苹果的便宜一些和戴尔的差不多。从这个角度看或许可以回答联想包不
包括IBM.既然thinkpad 基本上只能从官网买而官网质保更便宜，那么参保square
trade的联想用户更可能是从比如bestbuy,马鬃等买ideapad的用户。当然这仅仅是推测
，或许07年的时候联想官方质保更贵。
我第一台笔记本是ASUS，当时就瞅着其坚如磐石的口号，结果不幸中了转轴门，... 阅读全帖 |
|
i****x
发帖数: 17565 | 34 你那个国内烂校学报的”科研“结论跟以下所有结论矛盾的时候，你说我们该信谁呢？
In 2006 a large Danish group's study about the connection between mobile
phone use and cancer incidence was published. It followed over 420,000
Danish citizens for 20 years and showed no increased risk of cancer.[21] A
2011 follow-up confirmed these findings.[22]
The following studies of long time exposure have been published:
The 13 nation INTERPHONE project – the largest study of its kind ever
undertaken – was published in 2011 and did not find a solid li... 阅读全帖 |
|
m******r
发帖数: 1033 | 35 最近闭门造车，不接电话，不回电邮，不上网， 死几百咧，造了个模型，先用线性逻
辑回归，试来试去，性能不理想，AUC大概63% . 这也没什么奇怪的，并不是给你一堆
数，就能造个模型出来。 反正试来试去，就这一个变量可用，, 假定为A, AUC = 63%
然后我就用xgboost, 我的妈，AUC一下上升到95%， 96%，97%， 98%， 因为编程太弱
，我的土方法是：一个变量一个变量试， 都是manual work, 每次只跑一个变量，记录
重要结果，保存在excel里。 最终结果是：仅用A变量，AUC = 95%, 在此基础上加上一
点别的变量， AUC 很快飞涨到97%, 98%
我知道这种基于树的模型容易过度拟合, 就特意找了好几年前的老数据（真实数据）测
试。 测试的AUC性能一点都不下降， 和原来的差距小于1%. 所以不能说是过度拟合。
现在问题来了，我想来想去不明白为什么这个变量用在xgb有这么高的AUC? 不明白xgb
施了什么法术？向业务部门也很难解释，做个简单的tabulation, 能依稀看出一些
trend (这到能说明 线性回归下此变量达到AUC = 63... 阅读全帖 |
|
n******g
发帖数: 2201 | 36 你的变量大概是target 的别名 比如用每分钟速度预测时速 当然很准
[在 magliner (magliner) 的大作中提到：]
:最近闭门造车，不接电话，不回电邮，不上网， 死几百咧，造了个模型，先用线性逻
:辑回归，试来试去，性能不理想，AUC大概63% . 这也没什么奇怪的，并不是给你一堆
:数，就能造个模型出来。 反正试来试去，就这一个变量可用，, 假定为A, AUC = 63%
:然后我就用xgboost, 我的妈，AUC一下上升到95%， 96%，97%， 98%， 因为编程太弱
:，我的土方法是：一个变量一个变量试， 都是manual work, 每次只跑一个变量，记
录重要结果，保存在excel里。 最终结果是：仅用A变量，AUC = 95%, 在此基础上加上
一点别的变量， AUC 很快飞涨到97%, 98%
:我知道这种基于树的模型容易过度拟合, 就特意找了好几年前的老数据（真实数据）
测试。 测试的AUC性能一点都不下降， 和原来的差距小于1%. 所以不能说是过度拟合。
:现在问题来了，我想来想去不明白为什么这个变量用在xgb有这么高的AUC? 不明白xg... 阅读全帖 |
|
w**********y
发帖数: 1691 | 37 I agree with casact.
As I know in my company, people make the judgment subjectively by their
market sense and whether the model fitting is meaningful and explainable. It
is similar when they do feature selection. They just try, try and try. Try
different categorizing, different models, linear or nonlinear, and finally
analyze based on the output tabulation and simply p-value.
They never use some advanced statistical methods. The way they utilize
smoothing methods (Spline, LOWESS..), Simulation ( |
|
s****l
发帖数: 10462 | 38 Median income varied widely when looked at solely by gender.
Male ... $77,000
Female ... $55,500
This significant gender gap narrows considerably when the data
are cross-tabulated by level of education, length of experience,
primary specialty, etc.
Considerable differences were found when the data were analyzed
by ethnic origin.
White/Caucasian (not of Hispanic origin) ... $69,781
Pacific Islander (small sample size) ... $67,500
Asian ... $58,700
Hispanic/Latino ... $55,255
Black/African origin |
|
s*****0
发帖数: 357 | 39 周末杂事比较多，小孩的playdate，还有和朋友约定的网战等等，未能及时更新，见谅。
感谢楼上hbsr2010的一些概念更正，平时理论接触的少了，记忆有偏差，因为在网上随
便写些，也懒得查书，写的时候随兴所至，没有太注意。我尽量让文笔轻快些，让读者
不至于厌烦。以后尽量会避免误导，如有不确实之处，请务必指正。先行谢过了，因为
有自己的一摊东西要收拾，不能像做科研那样严谨了。
前文提到的t test, one way ANOVA以及相关的nonparametric都只有涉及到一个
variable，比如作对照实验，variable即treatment type，不同计量药物或者是
placebo。组和组的区分是由这个variable决定的。在涉及到更为复杂的模型前(比如
two way ANOVA, multiple regression)，我觉得还是先唠叨唠叨categorical data的
统计方法，毕竟做multiple regression之类的工作需要一定的统计背景，平时远没有
Chi-square这样的test用得多。所以先简后难了。
Categorical data在生... 阅读全帖 |
|
x*o
发帖数: 1037 | 40 Job description. OPT应该也可以，不过Postdoc不支持H1b
Qualifications:
•A Ph.D. in cell biology, molecular biology, or related field with a
minimum of 5 years of laboratory experience is required.
• A focus in the area of immunology and/or cell therapy is desired.
Previous Post-Doc experience is not required. Cell analysis skill sets such
as flow cytometry and microscopy are required. Experience with metabolomics
is preferred.
•General molecular biology and laboratory skills such as ELISA,
... 阅读全帖 |
|
