非常好的一本书，读了好几遍，每次都有新收获。只可惜我现在已经决定退坑了。

太香了，stat角度窥探ml的圣经，好香好香

Statistical Learning 最经典的入门教材。很多Machine Learning的书，没有关注太多模型背后的原理。但是要做好Machine Learning，这些必不可少。另，Youtube上可以搜到Hastie和小伙伴前些年的授课视频。

The linear model has been a mainstay of statistics for the past 30 years and remains one of our most important tools.

LAR uses least squares directions in the active set of variables.
Lasso uses least square directions; if a variable crosses zero, it is removed from the active set.
Boosting uses non-negative least squares directions in the active set.

the bias of the 1-nearest-neighbor estimate is often low, but the variance is high.

Monte Carlo is an extremely bad method; it should be used only
when all alternative methods are worse

有一个关于Metropolis算法的故事，非常流行：一晚，Edward、Metropolis和Marshall在派对上讨论这个问题，在鸡尾酒餐巾纸上写出了这个闻名的算法。他们最终的论文之所以写上妻子的名字，是为了安抚被整晚的技术性讨论所烦扰的女人Arianna和Augusta

Bagging; or Bootstrap AGGregatING, is an extension of bootstrapping to classification and regression problems. The main idea is to sample with replacement from the training data so that we now have B training data sets, each having n′≤n observations. The machine-learning algorithm is trained on each of the B data sets to form a committee.

还是觉得对这本书相见恨晚，研一写那么都web app有毛用啊，就应该踏踏实实的多读书啊~ 还好去实习了，还好发现原来啥都不知道，还好坚持把这本书啃完了，虽然理解的较为粗陋，要不要去读个博呢......真苦恼~

Both k-nearest neighbors and least squares end up approximating conditional expectatios by averages.

However, with a 0 − 1 outcome, this computation simplifies. We order the predictor classes according to the proportion falling in outcome class 1. Then we split this predictor as if it were an ordered predictor.

The clever design of network Net-5, motivated by the fact that features of handwriting style should appear in more than one part of a digit, was the result of many years of experimentation.

Scenario 1:
 The training data in each class were generated from bivariate Gaussian distributions with uncorrelated components and different means.

- Scenario 2:
The training data in each class came from a mixture of 10 low-variance Gaussian distributions, with individual means themselves distributed as Gaussian.

To get an ida of why, note that if the neighborhoods were nonoverlapping, there would be N/k neighborhoods and we would fit one parameter (a mean) in each neighborhood.

A retrospective sample of males in a heart-disease high-risk region
of the Western Cape, South Africa. There are roughly two controls per
case of CHD. Many of the CHD positive men have undergone blood
pressure reduction treatment and other programs to reduce their risk
factors after their CHD event. In some cases the measurements were
made after these treatments. These data are taken from a larger
dataset, described in  Rousseauw et al, 1983, South African Medical
Journal.

sbp		systolic blood pressure
tobacco		cumulative tobacco (kg)
ldl		low densiity lipoprotein cholesterol
adiposity
famhist		family history of heart disease (Present, Absent)
typea		type-A behavior
obesity
alcohol		current alcohol consumption
age		age at onset
chd		response, coronary heart disease

“In-sample error is not usually of direct interest since future values of the features are not likely to coincide with their training set value.But for comparison between models,in-sample error is convenient and often leads to effective model selection”

"over-all five or ten fold cross-validation are recommended as a good compromise"

I take it that the question is about LDA and linear (not logistic) regression.

There is a considerable and meaningful relation between linear regression and linear discriminant analysis. In case the DV consisting just of 2 groups the two analyses are actually identical. Despite that computations are different and the results - regression and discriminant coefficients - are not the same, they are exactly proportional to each other.

Now for the more-than-two-groups situation. First, let us state that LDA (its extraction, not classification stage) is equivalent (linearly related results) to canonical correlation analysis if you turn the grouping DV into a set of dummy variables (with one redundant of them droped out) and do canonical analysis with sets "IVs" and "dummies". Canonical variate...

在R. A. Fisher提出的著名的艾里斯(Iris)判别例子中……

Although the individual class density estimates may be biased, this bias might not hurt the posterior probabilities as much, esp near the decision regions. In fact the problem may be able to withstand considerable bias for the savings in variances such as naive assumption earns.

The clever design of network Net-5, motivated by the fact that features of handwriting style should appear in more than one part of a digit, was the result of many person years of experimentation.

This example also shows that neural networks are not a fully automatic tool, as they are sometimes advertised. As with all statistical models, subject matter knowledge can and should be used to improve their performance.

个人觉得“机器学习 --  从入门到精通”可以作为这本书的副标题。  机器学习、数据挖掘或者模式识别领域有几本非常流行的教材，比如Duda的模式分类，Bishop的PRML。Duda的书第一版是模式识别的奠基之作，现在大家谈论得是第二版，因为内容相对简单，非常流行，但对近20年取得统...

机器学习 --  从入门到精通

评论最下面的部分Version 1是我开始读这本书的时候写的东西，现在加上点基础部分。 对linear algebra, probability 要有非常强的直观认识，对这两个基础学的非常通透。Linear algebra 有几种常用的分解QR, eigendecomposition, SVD，搞清楚它们的作用和几何意义。Bayesian meth...

不是书评

我导师(stanford博士毕业)非常欣赏这本书，并把它作为我博士资格考试的参考教材之一。  感谢  ZHENHUI LI 提供的信息。本书作者已经将第二版的电子书放到网上，大家可以免费下载。 http://www-stat.stanford.edu/~tibs/ElemStatLearn/  网上还有一份solution manual, 但是似乎...

机器学习方面非常优秀的一本书

有人给我推荐这本书的时候说，有了这本书，就不再需要其他的机器学习教材了。 入手这本书的接下来两个月，我与教材中艰深的统计推断、矩阵、数值算法、凸优化等数学知识展开艰苦的斗争。于是我明白了何谓”不需要其他的机器学习教材“：准确地说，是其他的教材都不需要了；一本...

《统计学习基础》读后感：一点儿都不基础

这篇书评可能有关键情节透露

                        http://www-stat.stanford.edu/~tibs/ElemStatLearn/  还是老外人品好，把自己的书免费放在互联网上。

告诉大家一个下载此书的地方

这篇书评可能有关键情节透露

                        　　Elements of statistic learning is one of the most important textbooks on algorithm analysis in the field of machine learning. The authors of this book, Trevor Hastie, Robert Tibshirani and Jerome Friedman, are pioneers in the area and have done really b...

Classic!

读 ESL 快半年了，也读了差不多1/3，写个短评记录一下，等读完的时候再来改吧。然后简单对比下基本常见的机器学习教材。  我本科是学物理的，对于统计甚至概率论可以说是一无所知。入门的时候读的是周志华老师的《机器学习》，不过并没有读完的。一方面在家看书效率太低；另一...

半程记录

douban评论非要给出评价才能发表，这非常难决断 说你好呢，翻译的乱七八糟 说你不好呢，内容实在深刻  说起翻译来，这可是把中文说的比外文还难懂 Jiawei Han的数据挖掘让范明译的污七八糟 结果还让他来翻译这部经典，怀疑他在用google翻译  最后还是忍不住去图书馆复印了原版...

某人能不能别再毁书不倦了

中文翻译版大概是用google翻译翻的，然后排版一下，就出版了。所以中文翻译版中，每个单词翻译是对的，但一句话连起来却怎么也看不懂。最佳阅读方式是，看英文版，个别单词不认识的话，再看中文版对应的那个词。但如果英文版整个句子都不懂的话，那只有去借助baidu/google，并...

千万别看中文翻译版

这个简单的书评只是我个人的观点，所以我觉得先了解一下我的背景是有帮助的：本科计算机，数学功底尚可，研究生方向机器学习、数据挖掘相关应用研究。  缺点： 1，阅读此书前，读者需要具备基本的统计学知识，所以书的内容并不“基础”。 2，书中很少涉及到公式推导，细节并不...

机器学习、数据挖掘、模式识别的统计学观点

读了一个月，还在前四章深耕，在此说明一下，网上的 solution，笔记啊，我见到的，只有一个份做的最详细，准确度最高，其余的都是滥竽充数，过程推导乱来，想当然，因为该书的符号有点混乱，所以建议阅读该书的人把前面的 Notation 读清楚，比如书中 X 出现的有好几种形式，每...

此书不适合初学者，以及此书没那么神

对于新手来说，这本书和PRML比起来差太远，新手强烈建议去读PRML，接下来再看这本书。。我就举个最简单的例子吧，这本书的第二章overview of supervised learning和PRML的introduction差太远了。。。。读这本书的overview如果读者没有基础几乎不知所云。。但是PRML通过一个例子...

新手你千万别读这本书

非常难，一点都不element，是本百科全书式的读物，如果是初学者，不建议读 很多章节也没有细节，概述性的东西，能看懂几章就很不错了 其实每章都可以写成一本书，都可以做很多篇的论文 全部读懂非常非常难，倒是作为用到哪个部分作为参考资料查查很不错

难得要死

http://www-stat.stanford.edu/~hastie/local.ftp/Springer/ESLII_print3.pdf

感谢作者公开分享电子版

这篇书评可能有关键情节透露

                        Forecasting is always hard. It is not an easy task to capture the relationship between response variables and explanatory variables. Also, it is challenging to find a balance between bias and variance. Here, we introduce some methods, including time series ...

Forecasting with Machine Learning

一句话：本书是国外俗称的黄皮书（圣经）统计学系和机器学习的入门九阴真经宝典。好了，开始说自己的感想，我们墨大学校上课用的内容。斯坦福的巨匠们把统计学习这门“技术”深入浅出地娓娓道来，一手把我领进Machine Learning的殿堂。不具体展开内容了，统计学习是一门博大精...

A must read book in your life

原书真的是好书，毁在翻译上了。想恰饭可以翻其他的书，没必要荼毒这本书。又叹又恨。 原书真的是好书，毁在翻译上了。想恰饭可以翻其他的书，没必要荼毒这本书。又叹又恨。 原书真的是好书，毁在翻译上了。想恰饭可以翻其他的书，没必要荼毒这本书。又叹又恨。 原书真的是好书...

痛恨用Google翻译的罪人

这篇书评可能有关键情节透露

                        https://www.wenzhangba.com/duhougan/201712/195463.html 原文链接如下：看到下面这位读者的评论后，感谢前辈有态度的分享。 《The Elements of Statistical Learning》是一本由Trevor Hastie / Robert Tibshiran著作，Springer出版的Hardcover图书，本书定价：USD 89.95，页...

别人对英文版的一些读后感

Springer Series in Statistics 影印版

（英文版）

这本书在读者口中反复被冠以「圣经」一词，但它真正不可替代之处，不在于罗列了多少流行算法，而在于它坚持用统计学的视角重新审视机器学习：神经网络、支持向量机、分类树、提升法为何有效、又为何失效？短评里有人读完后「相见恨晚」，感叹课程与实习都没教给他的东西，这本书让他看清了自己「啥都不知道」；也有人啃完直呼「一点儿都不基础」——它不满足于告诉你怎么用模型，而是逼你理解模型背后的偏差与方差、估计的稳定性。它适合那些不满足于「调包」、想要追问原理的人；也提醒只想速成的读者：这里的收获需要耐心与数学直觉。这或许正是它评分居高不下的原因。