Recently, I found a need to explore the Intel DAAL MKL for Data Science and was having difficulties finding the proper installations for a working environment in one location. After several days and trial and failures, I was able to reach a final installation process that I think would be beneficial to all the other Data Scientist enthusiast who are looking to get started with their Data Science adventures, utilizing Visual Studio Code or JupyterLab. Posted below are my recommended steps to get a working environment on Windows 10.
1. Download and install the latest version of Anaconda: https://www.anaconda.com/distribution/
2. Download and install the latest Graphviz installer from their Web site: https://graphviz.gitlab.io/download/ (in my case, version graphviz-2.38.msi was active)
a. Install graphviz msi for all users
b. Navigate to Environment Variables: https://t.ly/Gz359
c. Create a new environment path for the Graphviz (need two links to be added): (in my case it was: C:\Program Files (x86)\Graphviz2.38\bin\ and C:\Program Files (x86)\Graphviz2.38\bin\dot.exe
d. Close all command and environment windows
e. Open a new cmd window and test for the existence of Graphviz: c:\Users\MyDrive>dot -v
i. You should get a report of the version and other info (if it fails, check the environment path entry and possibly repair your Graphviz installation)
ii. Ctrl-C to close the report
iii. Close the cmd window
iv. Reboot your PC
3. Navigate to Anaconda Command Prompt as Administrator and remain in the (base) environment:
a. In Windows 10, Search for anaconda and select the anaconda command prompt:
i. Right-Click on it and select to run it as Administrator
b. Navigate to the root of the (base) environment:
i. cd\
c. Get a current list of existing environments:
i. conda env list
d. Remove any unwanted environments:
i. conda env remove -n OldenvironmentName
e. Create new desired environment for Intel Data Science (ids) with the most current Conda libraries, and supported Python version 3.x:
i. conda create -n ids python=3 numpy pandas seaborn matplotlib scikit-learn daal4py jupyterlab -y
f. Activate the new environment:
i. conda activate ids
g. Install Graphviz with pip:
i. pip install graphviz
h. Install python support for Graphviz:
i. conda install pydot python-graphviz -y
i. Check that dot is accessible via cmd prompt:
i. dot -v
j. Ctrl-C to close the report
k. Register the following for intel DataScience enhancements:
i. set USE_DAAL4PY_SKLEARN=YES
ii. python -c "import sklearn"
l. Reboot our PC
When you return to your Desktop, you will be ready to use your new environment for Data Science processes