机器学习路线图
Artificial intelligence, machine learning, deep learning, internet of things etc. is the buzzword from the last few years. If someone in the strategy, business or IT consulting, doesn’t know at least the basics of it then I guess they are pretty much redundant.
人工智能，机器学习，深度学习，物联网等是近几年的流行词。 如果战略，业务或IT咨询领域的人员至少不了解其基础知识，那么我认为他们几乎是多余的。
To keep up with the waves two years back, I decided to get acquainted with machine learning and deep learning.
为了跟上两年前的潮流，我决定结识机器学习和深度学习。
Many of us want to make a career change to machine learning/AI or come from non-science background hence find it challenging to gain the requisite knowledge. In this article, I am going to share my journey and evolution with resources which helped me to gain knowledge in this area. I hope it will provide insight and help you to structure your learning path.
我们中的许多人都想改变职业以改用机器学习/人工智能或来自非科学背景，因此发现获得必要的知识具有挑战性。 在本文中，我将与资源分享我的旅程和发展，这些资源帮助我获得了这一领域的知识。 我希望它能提供见解并帮助您构建学习路径。
My Challenges
我的挑战
I studied mechanical engineering 20 years ago. Engineering degree prepared me to think systematically in a rational way. I did advanced calculus and linear algebra in my engineering my I forgot most of the concepts over the years. It became evident very soon that I need to learn a number of things as a pre-requisite even before starting to learn ABC of the machine or deep learning.
我20年前学习了机械工程。 工程学位使我能够理性地系统思考。 我在工程学中做了高级微积分和线性代数，但多年来我忘记了大多数概念。 很快就变得很明显，即使在开始学习机器的ABC或深度学习之前，我也需要学习很多东西作为先决条件。
- Brush up the concepts in linear algebra, calculus and probability 复习线性代数，微积分和概率的概念
- Learn basic statistics 学习基本统计
- Select the primary language 选择主要语言
- Learning Map 学习地图
I didn’t have the energy to read the mathematics textbooks, hence to revise I watched the videos on linear algebra, calculus and probability on Khan Academy and MIT open courseware. I will especially recommend the lectures by Prof. Gilbert Strang.
我没有精力阅读数学教科书，因此要进行修订，我在Khan Academy和MIT开放课件上观看了有关线性代数，微积分和概率的视频。 我将特别推荐Gilbert Strang教授的讲座。
I zeroed down to python as my primary programming language because of the vast libraries and developer support.
由于广泛的库和开发人员支持，我将python作为我的主要编程语言。
Once I have revised the mathematical concepts, then I started with Python. I have done programming before with C language, and I was already aware of the concept of looping, conditional statement etc.
修改完数学概念后，便开始使用Python。 我之前使用C语言进行过编程，并且我已经知道循环，条件语句等概念。
I found the book “Python Crash Course: A Hands-On, Project-Based Introduction to Programming “ by Eric Matthes very helpful to get a grasp on syntax and basic concepts. The other book which I used to hone my basic python skill is “Automate the Boring Stuff with Python, 2nd Edition: Practical Programming for Total Beginners” by Al Sweigart.
我发现Eric Matthes的书“ Python崩溃课程：基于项目的动手实践入门 ”对理解语法和基本概念很有帮助。 我用它来磨练我的基本技能蟒其他本书是“ 自动化无聊的东西与Python，第2版：实用程序设计总初学者 ”的Al Sweigart。
After gaining a sound footing on python, I focussed on the trinity of python package for machine learning, i.e. NumPy, pandas and matplotlib/seaborn.
在掌握python的良好基础之后，我将重点放在用于机器学习的python软件包的三位一体上，即NumPy，pandas和matplotlib / seaborn。
All three packages official documentation are very well written and easy to understand even for beginners. I started by NumPy and then moved to Pandas and finally to matplotlib/seaborn to get an idea on data visualisation.
所有这三个软件包的官方文档都写得很好，即使对于初学者也很容易理解。 我由NumPy开始，然后移至Pandas，最后移至matplotlib / seaborn，以获取有关数据可视化的想法。
It took me around 2 months with approx.3 hours reading to come until this point. I have never learnt statistics before that point, and my entire knowledge of statistics was limited to mean median and mode.
到现在为止，我花了大约2个月的时间阅读了大约3个小时。 在此之前，我从未学习过统计学，而我对统计学的全部了解仅限于中位数和众数。
I found the book “The Art of Statistics: Learning from Data” by David Spiegelhalter very helpful. It not only helped me to know statistics in a non-mathematical way but also helped a lot in term of data science visualisation. It taught me the way we can communicate different messages based on the same data points based on varying data visualisation.
我发现David Spiegelhalter的书“ 统计的艺术：从数据中学习 ”非常有帮助。 它不仅帮助我以非数学的方式了解统计信息，而且在数据科学可视化方面也提供了很多帮助。 它教会了我如何基于不同的数据可视化基于相同的数据点传达不同消息的方式。
Armed with the conceptual understanding of basic statistics, I read the book “Practical Statistics for Data Scientists: 50+ Essential Concepts Using R and Python” to build programmatic hands-on knowledge.
凭借对基本统计的概念理解，我读了《 面向数据科学家的实用统计：使用R和Python的50多个基本概念 》一书，以建立编程的动手知识。
After three months of studying the pre-requisites finally, I was ready to start learning machine learning. If I can name one book which has accelerated my learning and explains the concepts from beginners to advanced level, then it is “Hands-on Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems” by Aurelien Geron.
在学习了三个月的前提条件之后，我准备开始学习机器学习。 如果我能命名一本可以促进学习并从入门到高级的概念的书，那么它就是“使用Scikit-Learn，Keras和TensorFlow进行动手机器学习：构建智能系统的概念，工具和技术”。 ”由Aurelien Geron撰写。
Once you follow the learning path as I mentioned above from python to pandas and matplotlib, then there is not a single line in this book which you cannot understand. The complete hands-on approach in this book without the mathematical jargon is very helpful for people who do not have advanced mathematical knowledge.
一旦您按照我上面提到的从python到pandas和matplotlib的学习路径进行学习，那么本书中就没有一行是您无法理解的。 本书中没有数学术语的完整动手方法对那些不具备高级数学知识的人非常有用。
It took me a total of six months to reach the intermediate level machine learning proficiency and another two months to understand the basics of deep learning.
我总共花了六个月的时间才能达到中级水平的机器学习水平，又花了两个月的时间来了解深度学习的基础知识。
Key takeaways and learning from my journey
关键要点和从我的旅途中学习
Machine learning and data science stands proficiency stands on the firm knowledge of the following areas
机器学习和数据科学精通以下方面的扎实知识
- At least intermediate level python programming knowledge 至少具有中级Python编程知识
Advanced level knowledge on Numpy, Pandas and Matplotlib/Seaborn Python libraries
关于Numpy ， Pandas和Matplotlib / Seaborn Python库的高级知识
Do not be on haste to start machine learning libraries and books without having a firm grasp in these libraries. Thorough knowledge in these areas will make your machine learning swift and enjoying.
如果不牢牢掌握这些库，请不要急于启动机器学习库和书籍。 在这些领域中透彻的知识将使您的机器学习Swift而愉快。
Start your machine learning with easier and well-known algorithms like linear regression. Do not try with a most complex less known algorithm to appear knowledgable and impress others.
通过简单易懂的算法(例如线性回归)开始您的机器学习。 不要尝试使用最复杂的鲜为人知的算法来使其看起来可学并打动他人。
Scikit-learn is a very powerful and versatile machine learning library. It is also very userfriendly from a beginner perspective and one of the most well-documented libraries available. I will strongly recommend to start with Scikit-learn and spend a considerable amount of time to get acquainted before moving to other libraries.
Scikit-learn是一个功能强大且用途广泛的机器学习库。 从初学者的角度来看，它也是非常用户友好的，并且是最有据可查的库之一。 我强烈建议您从Scikit学习开始，并花费大量时间来熟悉其他图书馆。
Get yourself acquainted with machine learning before dipping your toes into the deep learning world. Rushing to deep learning without having a good understanding of machine learning is not going to help in the long run.
在深入学习之前，先熟悉一下机器学习。 从长远来看，在没有对机器学习有充分了解的情况下进入深度学习不会有帮助。
Start with Keras library for deep learning and you can park the TensorFlow for later.
从Keras库开始进行深度学习，您可以将TensorFlow停放在以后。
Finally, make use of free learning resources available on Coursera and Edx on the machine and deep learning.
最后，在机器和深度学习上利用Coursera和Edx上的免费学习资源。
I hope readers who are staring their machine and deep learning journey gain some insight on the learning approach from my journey outlined in this article and would be able to structure their learning path.
我希望那些盯着机器和深度学习旅程的读者能够从本文概述的我的旅程中对学习方法有所了解，并能够构建他们的学习道路。
I am interested to know the learning maps, key resources and approach of other proficient and student of machine learning.
我很想知道其他精通机器学习的学生的学习地图，关键资源和方法。
翻译自: https://towardsdatascience.com/machine-learning-roadmap-my-journey-and-evolution-a76ae656969
机器学习路线图
698
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
