![<?echo $_SERVER['SERVER_NAME'];?>](/template/twentyseventeen/skin/images/header.jpg)
There are many python machine learning resources available online for free, where do you start? How to proceed? From zero to python machine learning God is only 7 steps.
Getting started, the two most difficult words in English, the first is often the hardest, when there are too many choices in this area, usually people are falling apart.
Where do you start?
This article aims to become a knowledgeable practitioner with 7 newcomers who do not take the lead in Python machine learning, using free materials and resources. The main goal of this outline is to help you choose the many options available. It can be determined that there are many, but which one Is it the best? Are they complementary? What is the best order to use the resources?
Further, I made the assumption that you are not the following experts:
Machine learning
Python
Any Python machine learning, scientific calculation, or data analysis library
He may help you to have a basic understanding of the first or second or both, but it is not required. It may be possible to make up for some extra time in the previous steps.
The first step is basic Python skills
If we are going to use Python for machine learning, it is crucial to have some basic knowledge of Python. Fortunately, because of its popularity as a widely used general-purpose programming language, it has been applied to scientific computing and machine learning. Tutorials are not difficult. Python experience and programming levels are generally the focus of choosing a starting point.
First, you need to install Python. Because there are times when we need to use scientific computing and machine learning libraries, I suggest you install Anaconda. This is a Python implementation with industrial strength for Linux, OSX and Windows, which is required for machine learning installation. Packages, including Numpy, scikit-learn and Matplotlib (this is the author's point of view, I think pandas, scipy package is also essential). It also contains Ipython notebook, an interactive environment. I recommend using Python2.7 ( Translator's words: When the world is 3.X, then change Python3), it is still the dominant installation version for no other reason.
If you don't have programming knowledge, my advice is to start with the free online book below and then learn the material that follows.
Python The Hard Way by Zed A. Shaw
If you have become an experience, but not Python, or you just learned Python, I suggest learning one or all of the following:
Google Developers Python Course (highly recommended by visual learners, five-star rating)
An IntroducTIon to Python for ScienTIfic CompuTIng (from UCSB Engineering) by M. Scott Shell (Amazing Python Science Introduction, 60 pages)
For those who are looking for Python 30 minutes crash course, you can go (a treasure):
Learn X in Y Minutes (X = Python)
Of course, if you have Python programming experience, you can skip this step. Even so, I suggest you continue reading the Python documentation.
The second step of basic machine learning skills
Kachn Lipets' founder Zachary Lipton pointed out that people's understanding of data scientists is very different. It is actually a reflection in the field of machine learning, because the work of data scientists involves the use of machine learning algorithms to diversify. Is it necessary to have a deep understanding of the algorithm, which is more effective? Create SVM machine model and get information from it? Of course not, just like almost everything in life, the depth of theoretical understanding is relative to the actual application (this is the author's statement ~, the translator is not supported, do not understand you How to adjust the parameters and optimize the model? Ha ~). A deep understanding of the machine learning algorithm is beyond the scope of this book, generally requires a lot of time to invest in academics, or through high-intensity self-study.
The good news is that you don't need to have an understanding of the machine learning theory of PhD level in order to practice. Not all code farmers need theoretical computer science education for effective coding. These two points are the same (so I translated It’s a code farmer. Yes, I’m just spit it~ myself. :( ).
Andrew Ng's Course course is often given a five-star rating. But my advice is to browse the course notes that students have prepared before the online course. Skip notes on Octave (similar to Matlab language, not related to the Python we pursue. Translator's Note, from use In terms of words, I can feel the author's dislike of Matlab.) But be aware that these are not official notes, but it seems to capture the content of Andrew Ng course materials. Of course, if you have time and interest, you can participate in Andrew Ng Coursera. Machine learning course.
Unofficial Andrew Ng course notes
If you like various video lectures, you can watch Tom Mitchell's speech video. Below is his recent speech video. I (not me) feel that he is very approachable, I am his brain powder (no mistake, online popular) The machine learning video of Taiwan National University is modeled after Tom Mitchell. The logo is almost added to the upper right corner of the lower left corner. Yes, I am vomiting again.)
Tom Mitchell Machine Learning Lectures
Here, you don't need all the notes and videos. An effective strategy includes practical exercises, refer to the above notes and videos as appropriate. For example, when you encounter a regression model is a reality, read the regression section of Ng notes or watch MTIchell's return video. .
The third step is the science Python library probability
Ok. With a bit of Python programming experience and machine learning understanding. In addition to Python there are many open source libraries for handling machine learning practices. Usually, these are the main Python libraries used to perform basic machine learning tasks.
Numpy - N-dimensional array is very useful
Pandas - Python data analysis library, including structures such as dataframes
Matplotlib - 2D drawing gallery to generate publishing quality images
Scikit-learn - Machine learning algorithm for data analysis and data mining tasks.
The excellent materials for learning these are as follows:
Scipy Lecture Notes by Gaël Varoquaux, Emmanuelle Gouillart, and Olav Vahtras
This pandas tutorial is great:
10 Minutes to Pandas
Suizhou simi intelligent technology development co., LTD , https://www.msmsmart.com