x********r
发帖数: 1206 | 1 Scorecard 是 分卡
Metrics 就是 Measurement， 度量尺度。
就是 Metrics 是你定的，怎么量，怎么打分，然后打的分就是 Scorecard
一般在 Risk Management， Information Security， Financial Firm 里比较常见。 |
|
E**********e
发帖数: 1736 | 2 【 以下文字转载自 Statistics 讨论区 】
发信人: ExpressoLove (MoneyForNothing), 信区: Statistics
标 题: 做credit risk scorecard的朋友们， 请进来， 有问题求教
发信站: BBS 未名空间站 (Mon Feb 8 22:08:57 2016, 美东)
我现在在一个小的私人公司做risk modeling才半年多。前半年觉得自己做得很不错。
可是现在越觉得有很多问题很疑惑，现在抛出来，请有经验的大侠指导。
公司是做loan lending的小公司，比较新， 积累的charge off 数据4000不到， 这个
跟大银行动辄一两个million 的数据不一样。 modeling的数据不是很好。我就不自爆
奇丑了，主要表现是training 和test的AUC差别很大， 有很大overfitting。
现在问题来了。假设数据分成三个部分，数据一是training， 数据二是test， 数据三
是holdout。holdout 类似于future data， 用来测试最后model 表现。所以这部分数
据只... 阅读全帖 |
|
E**********e
发帖数: 1736 | 3 【 以下文字转载自 Statistics 讨论区 】
发信人: ExpressoLove (MoneyForNothing), 信区: Statistics
标 题: 做credit risk scorecard的朋友们， 请进来， 有问题求教
发信站: BBS 未名空间站 (Mon Feb 8 22:08:57 2016, 美东)
我现在在一个小的私人公司做risk modeling才半年多。前半年觉得自己做得很不错。
可是现在越觉得有很多问题很疑惑，现在抛出来，请有经验的大侠指导。
公司是做loan lending的小公司，比较新， 积累的charge off 数据4000不到， 这个
跟大银行动辄一两个million 的数据不一样。 modeling的数据不是很好。我就不自爆
奇丑了，主要表现是training 和test的AUC差别很大， 有很大overfitting。
现在问题来了。假设数据分成三个部分，数据一是training， 数据二是test， 数据三
是holdout。holdout 类似于future data， 用来测试最后model 表现。所以这部分数
据只... 阅读全帖 |
|
|
s**********9
发帖数: 1238 | 5 Work with PMs to maintain Scorecard metrics
职位需求上有这么一句
放狗查了一下，似乎是PROJECT MANAGEMENT的术语，哪位筒子能简单点说说这个系统是
干嘛的啊？ |
|
c********t
发帖数: 4527 | 6 IPO Scorecard: Update On Selected Initial Stock Offerings 05/29 02:36
PM
Company Symbol Percent Current Offer Date
Changyou.com Limited CYOU 104% $32.59 $16.00 3-Apr-09
Rosetta Stone Inc. RST 31% $23.57 $18.00 15-Apr-09
Mead Johnson Nutrition Co. MJN 26% $30.34 $24.00 10-Feb-09
SolarWinds Inc SWI 20% $15.06 $12.50 May 20-09
Bridgepoint Education Inc. BPI 17% $12.25 $10.50 |
|
W***n
发帖数: 11530 | 7 A Third-Quarter Mutual Fund Scorecard
By Christine Benz | 09-30-15 | 06:00 AM | Email Article
After remarkably smooth sailing for six-plus years, many investors had been
bracing themselves for volatility. They got it in the third quarter of 2015,
when a broad swath of investment types--especially stocks--took a tumble.
Worries over global growth, especially in emerging markets, was the key
catalyst for the sell-off. Although stocks had started to look shaky earlier
in the quarter, the Federal Re... 阅读全帖 |
|
|
l**********e
发帖数: 459 | 9
Golfshot GPS on iphone
这是我找过的最好用的，除了给你提供gps功能，记分外，每次打完就自动给你生成
scorecard。 还能够统计你打过的round，给你画表，画曲线走势。 |
|
|
|
A*f
发帖数: 3067 | 12 i dont know there is any standard method around.
but folks usually take a picture of scorecard and post the picture
i assume you know how to post a picture, follow "点击添加附件" |
|
e******e
发帖数: 2530 | 13 数字越大越难。
纸版的scorecard上面一般会有，不同开球的tee对应不同的难度，你上那个高尔夫球场
网站
，一般能找到那个数值。 |
|
M*****t
发帖数: 85 | 14 在看一个job description，某bank（非credit card），没有详细写明需要什么model，只说是做credit risk scorecard / rating model，而且用的编程工具是matlab或者sas。要计算probability of default和loss given default等metrics。我觉得这个job应该涉及的是retail credit risk，和Basel II有关，model应该是regression，decision tree等statistics model。
有可能是涉及corporate bonds方面的credit risk，或者用merton model吗？应该重点准备哪些modeling方法呢？大家给点意见，谢谢~~ |
|
A*****s
发帖数: 13748 | 15 算PD基本两类model
要么是reduced form，也就是regression based
要么是structural model，Merton算一种
不过Merton那个模型太糙了，我正在做的research就是要beat之
关于Structural model你可以看看Moody's KMV的研究文献，和他们的VK model
但是市面上我见过的model都不能推recovery，我倒是想了个办法但是不告诉你 :P
学术界的structural model一大堆，基本都是垃圾
reduced form不是我的茶，所以也没啥了解。
model，只说是做credit risk scorecard / rating model，而且用的编程工具是
matlab或者sas。要计算probability of default和loss given default等metrics。我
觉得这个job应该涉及的是retail cre
点准备哪些modeling方法呢？大家给点意见，谢谢~~ |
|
M*****t
发帖数: 85 | 16 在看一个job description，某bank（非credit card），没有详细写明需要什么model，只说是做credit risk scorecard / rating model，而且用的编程工具是matlab或者sas。要计算probability of default和loss given default等metrics。我觉得这个job应该涉及的是retail credit risk，和Basel II有关，model应该是regression，decision tree等statistics model。
有可能是涉及corporate bonds方面的credit risk，或者用merton model吗？应该重点准备哪些modeling方法呢？大家给点意见，谢谢~~ |
|
l***a
发帖数: 12410 | 17 basel II是和retail credit risk有关？我一直以为这是corporate risk的范畴。。。
哪位给科个谱？
model，只说是做credit risk scorecard / rating model，而且用的编程工具是
matlab或者sas。要计算probability of default和loss given default等metrics。我
觉得这个job应该涉及的是retail cre
点准备哪些modeling方法呢？大家给点意见，谢谢~~ |
|
H*******s
发帖数: 382 | 18 假定最后进入scorecard有两个变量，第一个变量是表示房产的，两个attribute,分别是
有与没有房产，第二个变量是occupation,几个attribute,其中一个attribute是retire
d, 最低bad rate，如果这个attribute 90%都是有房产的，那么这两个变量在Scorecar
d综合考虑，会不会double counted?
也就是说，retired因为有房产了，所以第一个变量拿了高分，第二个变量也是因为有了
房产而再次拿了高分？
谢谢了！ |
|
m******e
发帖数: 89 | 19 有谁看过Credit Risk Scorecards: Development and Implementation Using SAS
ＢＹ Mamdouh Refaat
是否值得买。103$
谢谢！ |
|
m******e
发帖数: 89 | 20 有谁看过Credit Risk Scorecards: Development and Implementation Using SAS
ＢＹ Mamdouh Refaat
是否值得买。103$
谢谢！ |
|
s******5
发帖数: 513 | 21 你好，你能给我一个书的copy么？
Credit Risk Scorecards: Development and Implementation Using SAS
ＢＹ Mamdouh Refaat
多谢
s********[email protected]
谢谢 |
|
f*******n
发帖数: 2665 | 22 我一直用SAS,但SAS里没有SVM, 所以我想用R中的SVM做个predictive model. 如果结果
好, 有没有可能建成scorecard并用在其他平台上?因为我们需要real time scoring,
而且我们不可能用R做scoring. |
|
|
E**********e
发帖数: 1736 | 24 我现在在一个小的私人公司做risk modeling才半年多。前半年觉得自己做得很不错。
可是现在越觉得有很多问题很疑惑，现在抛出来，请有经验的大侠指导。
公司是做loan lending的小公司，比较新， 积累的charge off 数据4000不到， 这个
跟大银行动辄一两个million 的数据不一样。 modeling的数据不是很好。我就不自爆
奇丑了，主要表现是training 和test的AUC差别很大， 有很大overfitting。
现在问题来了。假设数据分成三个部分，数据一是training， 数据二是test， 数据三
是holdout。holdout 类似于future data， 用来测试最后model 表现。所以这部分数
据只能在建模完后才拿出来。建模前是绝对不是偷看，防止数据“泄露”到modeling
过程。
我的主要问题是怎么预先选初始变量。我原先理解就是用数据一和二，初选个100左右
的变量，很多modeling的书谈到bivariate analysis，算pvalue， spearman
correlation，还有是么clustering，等等。然... 阅读全帖 |
|
g****e
发帖数: 1829 | 25 你的scorecard是用tree么？我没做过，所以具体不是很清楚。知道high level一点具
体方法的话我还能给你一些其他的建议。比如regularization。
最最基本的，你可以用Sample1，2做best subset的k fold，或者repeated k fold。选
变量在第一组，测试在第二组。你说的不能用sample1，2选是你的理解有误。他应该说
的是不能用合起来的sample做univariate test，然后把自变量pool定死。
现实中，如果几百个变量就没法做best subset了，可能要用一些偷工减料的办法比如
forward step。具体还是要看方法。很多减小variance的办法都要落实到具体方法。 |
|
e***n
发帖数: 215 | 26 写在前面：
我是一个二岁癌症患者的妈妈，我可爱聪明淘气的儿子三个月前被诊断出横纹细胞肌肉
瘤。
目前在boston 就医，现在把就医过程中搜集的一些材料分享出来，供大家参考：）
希望每个孩子都能健康成长！
USNEWS 会对医院进行排名，这个是link。
http://health.usnews.com/best-hospitals
一下是儿科癌症医院排名：
#1 Cincinnati Children's Hospital Medical Center
Cincinnati, OH
See Cancer scorecard
100.0 / 100
#2 Dana-Farber Boston Children's Cancer and Blood Disorders Center
Boston, MA
See Cancer scorecard
99.2 / 100
#3 Children's Hospital of Philadelphia
Philadelphia, PA
See Cancer scorecard
95.4 / 100
#4 Children's Hos... 阅读全帖 |
|
e***n
发帖数: 215 | 27 经常看到有人想来美国看病，但不知道哪个医院比较好。
USNEWS虽然不是很权威，但至少提供了ranking
出处：http://health.usnews.com/best-hospitals/rankings/cancer
#1 University of Texas MD Anderson Cancer Center
Houston, TX
See Cancer scorecard
Ranked in Houston metro area and Texas
100.0 / 100
#2 Memorial Sloan-Kettering Cancer Center
New York, NY
See Cancer scorecard
Ranked in New York metro area and New York
93.0 / 100
#3 Mayo Clinic
Rochester, MN
See Cancer scorecard
Ranked in Minnesota
89.2 / 100
#4 Johns Hopkins Hospital
Baltimore,... 阅读全帖 |
|
A*******e
发帖数: 8 | 28 please send your resume to a**********[email protected],
and let us know which role you are interested in, when you can start working
, your current visa status, need sponsorship or not etc.
Role 1:
Optimization Specialist
9+ Month Contractor Role
Location: Stamford, CT
Industry: Travel and Lodging Services
Descriptions
Work closely with business teams and technical teams to gather
, analyze and understand business requirements, participate and contribute
to solution models, and design/d... 阅读全帖 |
|
A*******e
发帖数: 8 | 29 please send your resume to a**********[email protected],
and let us know which role you are interested in, when you can start working
, your current visa status, need sponsorship or not etc.
Role 1:
Optimization Specialist
9+ Month Contractor Role
Location: Stamford, CT
Industry: Travel and Lodging Services
Descriptions
Work closely with business teams and technical teams to gather
, analyze and understand business requirements, participate and contribute
to solution models, and design/d... 阅读全帖 |
|
A*******e
发帖数: 8 | 30 please send your resume to a**********[email protected],
and let us know which role you are interested in, when you can start working
, your current visa status, need sponsorship or not etc.
Role 1:
Optimization Specialist (Pay is very good)
9+ Month Contractor Role
Location: Stamford, CT
Industry: Travel and Lodging Services
Descriptions
Work closely with business teams and technical teams to gather
, analyze and understand business requirements, participate and contribute
to solution m... 阅读全帖 |
|
t*****9
发帖数: 33 | 31 印度人顾问公司，但好处是当data science第一人，和入职后直接跟大银行做，
project已谈好。现在就要人，有兴趣的朋友可以试试。MS/PhD 加四年经验。
-Need for a data scientist in NYC to build some scorecards for a lending
business
-The person should have knowledge of consumer lending, credit bureaus and
predictive modeling.
-Developing data-drives decision to build consumer credit origination/
underwriting scorecards.
· Work with the teams to build the credit scoring, pricing, and risk
scorecards
· Implement and test data-mining projects on our large data rep... 阅读全帖 |
|
A*******e
发帖数: 8 | 32 please send your resume to a**********[email protected],
and let us know which role you are interested in, when you can start working
, your current visa status, need sponsorship or not etc.
Role 1:
Optimization Specialist (Pay is very good)
9+ Month Contractor Role
Location: Stamford, CT
Industry: Travel and Lodging Services
Descriptions
Work closely with business teams and technical teams to gather
, analyze and understand business requirements, participate and contribute
to solution m... 阅读全帖 |
|
t*****9
发帖数: 33 | 33 印度人顾问公司，但好处是当lead data science第一人，和入职后直接跟大银行做 (
like GS,JPM,MS)，
project已谈好。现在就要人，有兴趣的朋友可以试试。MS/PhD 加四年经验。
-Need for a data scientist in NYC to build some scorecards for a lending
business
-The person should have knowledge of consumer lending, credit bureaus and
predictive modeling.
-Developing data-drives decision to build consumer credit origination/
underwriting scorecards.
· Work with the teams to build the credit scoring, pricing, and risk
scorecards
· Implement and test data-mining project... 阅读全帖 |
|
A*******e
发帖数: 8 | 34 please send your resume to a**********[email protected],
and let us know which role you are interested in, when you can start working
, your current visa status, need sponsorship or not etc.
Role 1 - Commercial Bank:
Full time (experience is essential)
Location: New York, NY
Industry: Commercial Bank
Develop and use credit scoring models and other risk analytics in consumer
lending.
Use sophisticated statistical techniques to develop scorecards, decision
tree based customer segmentation schemes, profi... 阅读全帖 |
|
t*****9
发帖数: 33 | 35 印度人顾问公司，但好处是当lead data science第一人，和入职后直接跟大银行做 (
like GS,JPM,MS)，
project已谈好。现在就要人，有兴趣的朋友可以试试。MS/PhD 加四年经验。
-Need for a data scientist in NYC to build some scorecards for a lending
business
-The person should have knowledge of consumer lending, credit bureaus and
predictive modeling.
-Developing data-drives decision to build consumer credit origination/
underwriting scorecards.
· Work with the teams to build the credit scoring, pricing, and risk
scorecards
· Implement and test data-mining project... 阅读全帖 |
|
m**c
发帖数: 7299 | 36 连环杀手--兰迪 史蒂芬 卡夫(Randy Steven Kraft)
姓名∶Randy Steven Kraft
外号∶记分卡杀手（Score-Card Killer）
出生日期∶1945年3月19日
现况∶在加利福尼亚州的圣昆廷州立监狱里服刑
杀人数目∶16+人,均为男性(也有人认为，受害人应至少67人)
杀人方法∶麻醉，捆绑，折磨
犯案地点∶美国加利福尼亚州(California)
杀人时期：1971年至1983年
被捕日期∶1983年5月14日
Randy Steven Kraft是一名美国连环杀手。虽然他被控谋杀了16人，但是他涉嫌谋杀了
至少67人。在1971-1983年期间，在加利福尼亚的高速公路上发生了很多命案。所有的
受害人都是年轻男子和少年，大多数受害人生前都遭受了折磨和X虐待。一些人还被焚
尸。许多受害人生前饮用了大量的酒和镇定剂。但直到1983年5月14日，Randy Steven
Kraft才被捕获。他还用密码为所有受害人编写了一个名单。在1988年，法庭判定他无
罪，但是他自己供认了所有罪行，并被判死刑，但尚未行刑。他现在还在监狱的死亡名
单上。
World... 阅读全帖 |
|
t****z
发帖数: 8931 | 37 2015 Toronto World Rankings
1 50 The Economist's Best Place to Live
Toronto was named the best place to live in the world by The Economist
based on 6 of its indexes.
5 24 Scorecard on Prosperity Toronto fell from 3rd to 5th place in the
annual Toronto Board of Trade's Scorecard on Prosperity.
8 50 Safe Cities Index Toronto ranked 8th of 50 cities in a new urban safety
report by The Economist Intelligence Unit.
12 50 Sustainable Cities Index Toronto ranks 12th, the highest North
American cit... 阅读全帖 |
|
t****z
发帖数: 8931 | 38 http://www1.toronto.ca/wps/portal/contentonly?vgnextoid=9048692
2014 Toronto World Rankings
TorontoRanking NumberofCities Report Description
1 51 Focus on Tax KPMG ranks Toronto as world's most tax competitive major
city.
1 50 Resilient Cities Toronto was ranked the world's most resilient city by
the Grosvenor Group.
1 25 Youthful Cities Toronto voted most youthful city by young adults aged
15-29.
1 - Intelligent Community Forum Toronto was named the world's top
intelligent community of 2014... 阅读全帖 |
|
|
e****d
发帖数: 333 | 40 招人的就是我直接老板，所以如果合适的话会很快。
Tier 1 Wall Street company is looking for a seasoned financial engineer PhD.
to join our model development team within the global Model group.
Job Duties:
-Conduct research to enhance and improve rating models (quantitative models
and scorecards).
-Publish research, attend external research conferences and participate in
professional training for the analytical staff.
-Supervise research and model development projects.
-Develop and maintain models and scorecards used in t... 阅读全帖 |
|
b****d
发帖数: 35 | 41 Urgent Hire, expect to start ASAP.有兴趣请站内联系。
AVP, Quantitative Analyst (Credit Scoring Models)
This is a position within the Quantitative Development & Analysis Group.
The incumbent is expected to design, develop, implement and maintain various
credit scores and their applications in credit decisioning for commercial
lending businesses of small and mid-tickets. The scores are built by
industry standard statistical methods incorporating data from various
sources - external and internal data of Fir... 阅读全帖 |
|
b****d
发帖数: 35 | 42 Urgent Hire, expect to start ASAP.有兴趣请站内联系。
AVP, Quantitative Analyst (Credit Scoring Models)
This is a position within the Quantitative Development & Analysis Group.
The incumbent is expected to design, develop, implement and maintain various
credit scores and their applications in credit decisioning for commercial
lending businesses of small and mid-tickets. The scores are built by
industry standard statistical methods incorporating data from various
sources - external and internal data of Fir... 阅读全帖 |
|
|
|
b****d
发帖数: 35 | 45 【 以下文字转载自 JobHunting 讨论区 】
发信人: budbud (I am bored), 信区: JobHunting
标 题: job opening in northern NJ, key word: statistical modeling, SAS, commerial lending
发信站: BBS 未名空间站 (Wed Oct 16 20:26:11 2013, 美东)
Urgent Hire, expect to start ASAP.有兴趣请站内联系。
AVP, Quantitative Analyst (Credit Scoring Models)
This is a position within the Quantitative Development & Analysis Group.
The incumbent is expected to design, develop, implement and maintain various
credit scores and their applications in credit decisi... 阅读全帖 |
|
g******e
发帖数: 760 | 46 改scorecard就太不对了，不过对于Annika的事，就见仁见智了。技术上他还是不错的。
scorecard |
|
ET
发帖数: 10701 | 47 thanks. dude.
good advice for me.
Usually, i always rush to achieve something, which is usually not good for me.
Just like this time, say, i played 92 yesterday, then checked the scorecard
and thought whether i
could make up some strokes today.So I played again.
I marked some data on my scorecards. like the puttings I used for each hole.
the misshots i made. Then i can check where are my weekpoints.
Golf is fun. Really different from those sports i used to be good at, somehow.
and learning somet |
|
ET
发帖数: 10701 | 48 golf rule is quite "simple", 36 (something) total, but quite confusing. That
's why so many cases happened everywhere, and every tournaments.
basically, most of the penality after signing a wrong scorecard is
disqualification, add some stokes to the existing score cards before signing
a scorecard.
once game is done, all case is closed, no further penality on the given
result.
something like that. |
|
v*****r
发帖数: 2325 | 49 just out of curiosity
Tees Yardage Rating Slope Rating Slope
Black 6,935 73.7 139 - -
scorecard
http://www.hudsonhillsgolf.com/play/the-course/scorecard/
NO 225 par 3 or 505 par 5, either not this course or data not accurate...
ball fly over 225, it's going to hard to see clearly, that's what i found
out at range
场 76.1/146 for black，和BETHPAGE有一拼。所以觉得他的HOME COURSE应该也在附近吧
maintanence。 等我练好了，也上那场子打打 :P |
|
d******s
发帖数: 674 | 50 I have golfGPS from golfshot.
haven't used viewti but one of my friend has it.
I am happy with the golfshot GPS, bought it once it came out. usded
scorecard version from them and loved that app.
the GPS app functions include:
standard gps functions (yardages or meters)
handicap calculation, auto updating, handicap scoring
keep current and previous scorecards
statistics for current and previous rounds
Both apps have decent ratings and Golfshot GPS costs little more. |
|
