Skip to content
Machine learning

Set up your machine-learning environment

There are three ways to set up an environment in which to work on Machine Learning.

Docker command

  1. Install Docker
  2. Create a directory q and save your kc.lic (or k4.lic) and l64.zip files in it
  3. Run:
$ docker run --rm -it -v `pwd`/q:/tmp/q kxsys/embedpy
kx@1ba5d6c29709:~$ q
KDB+ 3.5 2017.11.08 Copyright (C) 1993-2017 Kx Systems
l64/ 4(16)core 7905MB kx 0123456789ab 172.17.0.2 EXPIRE 2018.12.04 bob@example.com KOD #0000000

q)

You can drop straight into q with:

$ docker run --rm -it -v `pwd`/q:/tmp/q kxsys/embedpy q
KDB+ 3.5 2017.11.08 Copyright (C) 1993-2017 Kx Systems
l64/ 4(16)core 7905MB kx 0123456789ab 172.17.0.2 EXPIRE 2018.12.04 bob@example.com KOD #0000000

q)

Lastly you can also pipe your program in:

$ echo 'p)print(1+2)' | docker run --rm -i -v `pwd`/q:/tmp/q kxsys/embedpy q -q
3

Alternative setup with JupyterQ

Install Docker. Create a directory q and place your kc.lic (or k4.lic) and l64.zip files in it.

Run 

docker run --rm -it -v `pwd`/q:/tmp/q -p 8888:8888 kxsys/jupyterq

Now point your browser at http://localhost:8888/notebooks/kdb%2BNotebooks.ipynb.

Build instructions for the image

Download via Anaconda

The three Kx packages can be downloaded from anaconda.org/kx:

  • kdb
  • embedpy
  • jupyterq

Currently available for Linux and macOS; soon to be available for Windows too.

They are in a dependency tree. If you install embedpy it will automatically install kdb. If you install jupyterq it will install embedpy and kdb.

The commands are as follows:

conda install -c kx kdb
conda install -c kx embedpy
conda install -c kx jupyterq

At present, the packages work only from the base environment.

Before starting q, please run the following commands:

source deactivate base
source activate base

When you first run q it will ask the following questions:

Please provide your email (requires validation):
Please provide your name:
If applicable please provide your company name (press enter for none):

This will then reach out to the Kx license server and generate a kc.lic.

This in turn sends an email confirmation link to validate the license file.

Do it yourself

  1. Install kdb+
  2. Install embedPy
  3. Install JupyterQ