Originally Published 12 May 2017 This page gives quick and easy steps to setup your machine for machine learning and deep learning. Hi guys, In python there are many prebuilt libraries which can be used for machine learning and deep . But one major problem lies in setting up the environment for development using these libraries (especially for Windows based machines). learning Since most of these libraries were built to be used on Linux (Ubuntu etc), its easier to setup the working environment in . At the same time its a very confusing and lengthy procedure on Windows machine. Linux So below are the steps to get you up and running for ml and deep learning development on Windows: Step 1: Download Anaconda Distribution For your version of system 32bit or 64bit from Its better to select the version for Python 3.6 or 3.5.x since Tensorflow is yet only available for Python 3.5. here Fig 1. Anaconda Navigator download page Step 2: Install Anaconda Follow through the on-screen instructions Next->Next->Next-> then : check the box to add python to environment variables path. also make Anaconda as the default python checkbox to maintain uniformity. Fig 2. Installation step for Anaconda 3 The following packages will be preinstalled in the Navigator: Spyder (IDE of choice) Jupyter Notebook (for interactive programming) Qt console ( for inline figures and graphics) Fig 3. Initial view of the Anaconda Navigator Panel Step 4: Now go to the Environments tab on the left side of the Navigator Here click on new environment button, and Create type in the Environment name you want (say environment) Check only the Python check box ( uncheck R) In the Python version tab select Python 3.5, since Tensorflow currently only works for 3.5 Fig 4. Environment selection and creation page Once your new environment for Python 3.5 has been created it would appear like this Fig 5. Installed packages in a given environment Now from the search packages box, type-in and download the following packages: Jupyter Scikit-learn qtconsole matplotlib numpy pandas pip scipy Step 5: Once all above packages are installed, open up a terminal from your newly created Environment. Fig 6. Steps to open a terminal in given environment Once the terminal opens up type in the following for installing Theano: conda install theano pygpu Let the download and installation finish. After this type in the following for installing Tensorflow: For CPU pip install --ignore-installed --upgrade [https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-1.1.0-cp35-cp35m-win_amd64.whl](https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-1.1.0-cp35-cp35m-win_amd64.whl) For GPU pip install --ignore-installed --upgrade [https://storage.googleapis.com/tensorflow/windows/gpu/tensorflow_gpu-1.1.0-cp35-cp35m-win_amd64.whl](https://storage.googleapis.com/tensorflow/windows/gpu/tensorflow_gpu-1.1.0-cp35-cp35m-win_amd64.whl) Once tensorflow has been successfully installed, type the following for installing Keras: pip install git+git://github.com/fchollet/keras.git Once everything has been installed,go to next Step. If not search for your specific error on Google, you will find numerous discussions and answers on StackOverflow. Otherwise ping me. Step 6: Testing Now to test whether scikit-learn, Theano , Tensorflow and Keras are working properly; open up a terminal just like the previous step but with python Fig 7. Opening a python shell in a given environment Once the terminal opens up, it should appear like this. Fig 8. Python Shell showing the installed version 3.5.3 Now type the following commands to test their working import tensorflow as tf import theano import keras If there are no errors, then Congratulations you have successfully setup your Environment for machine learning and deep learning. If some errors pops up, copy the error content as search for solution on Google and stack overflow. This will only help you later in fixing other such minor issue. Otherwise ping me :D Originally published at mandroid6.github.io on May 12, 2017. Checkout my other posts on machine learning and deep learning : https://medium.com/@razzormandar
