西格尔零点猜想_我从埃里克·西格尔学到的东西

西格尔零点猜想

I finished reading Eric Siegel’s Predictive Analytics. And I have to say it was an awesome read. How do I define an awesome or great book? A book that changes your attitude permanently. You must not be the same person that you were before you picked up the book. It impacts one or more aspects of your life: personal, financial, social, romantic, family, or professional. Also, I read a book only if I can use what I learn from it. I don’t read it just for the sake of learning something. It needs to be practically usable in one of the areas of my life immediately if a book has to be on my desk. So yes, this book made a big impact not only on my professional understanding of data science but it also helped me uncover my interests. There are two primary things to learn from the book. First one is from the five effects:

我读完了Eric Siegel的Predictive Analytics 。 我不得不说这是一本很棒的书。 我如何定义一本很棒的书？ 一本书可以永久改变您的态度。 您一定不能和拾起书之前的那个人一样。 它会影响您生活的一个或多个方面：个人，财务，社交，浪漫，家庭或职业。 另外，只有在我可以使用从书本中学到的知识的情况下，我才会读一本书。 我不只是为了学习一些东西而阅读它。 如果必须在我的书桌上放一本书，它必须立即在我的生活中的一个领域中可用。 因此，是的，这本书不仅对我对数据科学的专业理解产生了重大影响，而且还帮助我发现了自己的兴趣。 从这本书中有两点要学习。 第一个是来自五个效果 ：

1. The Prediction Effect
   预测效果
2. The Data Effect
   数据效果
3. The Induction Effect
   感应效应
4. The Ensemble Effect
   合奏效果
5. The Persuasion Effect
   说服力

Without giving away the book’s ideas, Eric has hidden a lot of his experience in predictive analytics into these effects. The prediction effect proves that a lesser accurate prediction is better than a guess in business. The data effect says the data always has a story to tell and there is always something valuable to learn from it. The induction effect proves that it is the art that drives machine learning. The ensemble effect explains how the concept of synergy is useful in prediction. The persuasion effect connects marketing techniques, business-sense, and A/B testing. You might think it is all so simple and you already know it and you might be right unless you have a decade of experience in predictive analytics. And you still can learn something new in the book. Each effect is explained with real-life business cases. The book is filled with practical business results obtained from applying these effects. The most contrasting thing is he was a professor in a university but his writing style is practical, non-academic, and business-results oriented.

埃里克(Eric)在不放弃本书思想的前提下，将他在预测分析中的许多经验隐藏在这些影响中。 预测效果证明，准确度较低的预测比业务中的猜测更好。 数据效应表明数据总是有故事要讲的，总有一些值得学习的东西。 归纳效应证明，这是驱动机器学习的艺术。 合奏效应解释了协同作用的概念如何在预测中有用。 说服效果将营销技术，业务感知和A / B测试联系起来。 您可能会认为它是如此简单，并且您已经知道了，并且可能是对的，除非您在预测分析方面拥有十年的经验。 而且您仍然可以学习这本书中的新内容。 每种效果均通过实际业务案例进行解释。 这本书包含了通过应用这些效果而获得的实际业务成果。 最相反的是他是大学的教授，但是他的写作风格是务实，非学术性和商业成果导向的。

From Eric Siegel’s Book

埃里克·西格尔的书

The second thing I learned from the book is the understanding of the subject itself. I have taken several data science courses and written short programs using Pandas, NumPy, and scikit-learn. I have built a few machine learning models and I thought I knew something. I was wrong. This book taught me the usefulness of machine learning in real-life. Writing code to build, test, and evaluate models is not understanding machine learning. This book gives a detailed explanation of what machine learning is. There is an even more detailed explanation of decision trees without a single line of code. That in itself shows the grip of Eric’s understanding of machine learning modeling. Then there is a good amount of coverage of important topics like correlation does not imply causation, how models over-learn, and why training-test data split exists. Of course, all of it with real-life business cases. What looks like business risk, Eric converts it into an opportunity using predictive analytics. There is not a page in the book where he loses the focus from using predictive analytics to solve business problems.

我从这本书中学到的第二件事是对主题本身的理解。 我参加了一些数据科学课程，并使用Pandas，NumPy和scikit-learn编写了简短的程序。 我建立了一些机器学习模型，我以为我知道一些。 我错了。 这本书教会了我机器学习在现实生活中的有用性。 编写代码以构建，测试和评估模型并不能理解机器学习。 本书详细介绍了什么是机器学习。 无需一行代码，就可以更详细地解释决策树。 这本身就表明了Eric对机器学习建模的理解。 然后，对重要主题的讨论很多，例如相关性并不意味着因果关系 ， 模型如何过度学习以及为何存在训练测试数据拆分 。 当然，所有这些都与真实的业务案例有关。 看起来像业务风险的Eric使用预测分析将其转换为机会。 书中没有一页他会因为使用预测分析来解决业务问题而失去了重点。

My professional interests have permanently changed after reading the book. Now I am curious and very much interested in learning and finding more about how machine learning uncovers financial frauds, how a machine learning program can be applied to marketing or advertising problem in a business, and how it can be used in law enforcement. All of which I was least interested in doing before reading the book. I skipped some parts of the book but still, it was a mind-bending experience.

读完这本书后，我的专业兴趣永久地改变了。 现在，我对学习和发现更多有关机器学习如何发现财务欺诈，如何将机器学习程序应用于企业中的营销或广告问题以及如何在执法中使用的信息感到非常好奇和非常感兴趣。 在阅读本书之前，我最不感兴趣的是所有这些。 我略过了本书的某些部分，但那仍然是一种令人难以置信的经历。

翻译自: https://towardsdatascience.com/what-i-learned-from-eric-siegel-1399e1e6d944

西格尔零点猜想