[Definitive Edition] Building an environment for learning "machine learning" using Python on Windows

Introduction

Many people said, "I want to study machine learning and artificial intelligence, but I'm stumbling on the environment construction before learning programming **", and I wrote an article ** to never fail in the environment construction. It was decided to do.

Currently, "** Machine Learning / Artificial Intelligence" De-Black Box "Seminar **" is being held in Tokyo, and its students We ask all of you to build a Python environment on your PC in advance. However, there were large individual differences depending on the PC, and some people failed to install. Therefore, I searched for a method that anyone could set up well, came to the conclusion of the installation method of this article, and now there are almost no people who stumble on the installation even at the seminar of about 30 people, and everyone can concentrate on the seminar. .. Therefore, I think that this article is probably the most stable installation article at the moment (2017.03.18) on Windows PC (recommended is Windows 10).

There is more than one way to build a Python machine learning environment on Windows, and you can try installing it in various ways. It's nice to have multiple methods because even if you make a mistake, you can try other methods, but ** it can be a cause for beginners to wonder which method to build the environment **.

Therefore, in this article, we are targeting the following people.

• ** First-time Python users **
• Python is not a project but ** people who use it for themselves ** (in short, people who do not need a virtual environment)

Please refer to this article to build an environment that will not fail.

Thank you for your cooperation!

If you find this article helpful, I would be grateful if you could like this article.

About the author

I am ** Kikagaku Co., Ltd. ** Representative Director ** Ryosuke Yoshizaki ** My name is. Currently, "** Machine Learning / Artificial Intelligence De-Black Box Seminar **" and "** Machine Learning Online Tutor ** ”is operated.

Biography

Kikagaku Co., Ltd.

Providing educational services for machine learning and artificial intelligence

We look forward to your follow-up

We provide information on machine learning and artificial intelligence from a business perspective and recommended reference books.

President and CEO Ryosuke Yoshizaki Twitter：@yoshizaki_kkgk Facebook：@ryosuke.yoshizaki Blog: Blog of Kikagaku representative

Now that the introduction is long, let's start building the environment!

Install Python and required libraries

First, go to the ** here ** page, download the ** Anaconda ** distribution file, and install it.

• If you have already installed Python, it is safe to uninstall the already installed Python (recommended).

Anaconda will do everything from building a Python environment to installing some necessary libraries at once.

It will take some time to download.

After downloading, double-click the downloaded file to execute the installation.

Basically it is OK, but if you want to specify the location of the download file somewhere, change this location to your favorite location.

• If you are not particular about it, you can leave it as it is.

Leave the checks for "Add PATH (top)" and "Set Python installed with Anaconda as default (bottom)" to install.

Installation will be completed in a few minutes.

Why not use the official Python installation file?

Of course, you can also download the installation file from the official Python page and install it.

However, with this method, when installing the library required to install the machine learning library scikit-learn called numpy or scipy depending on the PC," there is no BLAS or ATLAS "" C I often encountered difficult-to-handle errors such as "no compiler". It is very difficult to solve this, and ** Anaconda ** builds this environment in one shot, so no one has failed in building the environment.

Make sure Python and libraries are installed

Open "Windows PowerShell" (* you can also use a command prompt) that comes standard with Windows.

Enter the following command on this screen, and if it works correctly as shown below, the installation is complete.

Operation check

$ python
>>> import numpy         #Library for linear algebra
>>> import scipy         #Computer algebra library
>>> import matplotlib    #Visualization library
>>> import pandas        #Data processing library
>>> import sklearn       #Library for machine learning
>>> exit()               #Exit Python interactive mode

If there are no particular errors here, the installation is complete.

Install the library for deep learning

Open "Windows PowerShell" (* you can also use a command prompt) that comes standard with Windows.

Enter the following command on this screen to install "** Chainer **", a Python library for deep learning.

Install chainer

$ pip install chainer

pip is a very useful tool for managing Python libraries such as installing and uninstalling them. It's used a lot, so it's a good idea to remember how to use it.

Operation check

Make sure you have chainer installed using the same steps as before.

Operation check of chainer

$ python
>>> import chainer  #Loading a library for deep learning
>>> exit()          #End of interactive mode

Installation is complete if there are no particular errors as described above.

in conclusion

Thank you for building the environment on Windows. Were you able to build the environment smoothly?

I used Anaconda to install it in a very neat state, so when I needed another library in the future, I didn't see pip or this time, but if I use the conda command, Rest assured that the additional libraries will generally install successfully.

We are very much looking forward to learning machine learning and revolutionizing various industries, and we hope that this article has helped you. If you found this article useful, we would appreciate it if you could like the article.

We look forward to your follow-up

We provide information on machine learning and artificial intelligence from a business perspective and recommended reference books.

President and CEO Ryosuke Yoshizaki Twitter：@yoshizaki_kkgk Facebook：@ryosuke.yoshizaki Blog: Blog of Kikagaku representative

Until the end Thank you for reading.