This is a demonstration of a range of interactive visualisations for data analysis. The notebook is available on mybinder which allows for interacting with the visualisations. This is the fastest way to get started.

Installation of the dependencies is through conda which can be obtained by installing miniconda.

Once conda is installed, all dependencies can be installed with the command

conda env create

which will create the interactive-visualisation environment based on the dependencies specified in the environment.yml file.

Once the dependencies are installed, running the command

conda activate interactive-visualisation

will load all the packages ready for use.

The data files which are used in the demos need to be downloaded before this notebook can be run. These data files can be downloaded with the command

make

The notebook interactive.ipynb is designed to work as a jupyter notebook and is not compatible with juptyerlab without some tweaking. Running the command

jupyter notebook

will open a jupyter notebook interface in the current directory.

The interactive.ipynb notebook is designed to be presented as slides, however for brevity and clarity much of the code is in hidden cells. So to properly run this as an interactive presentation all the cells need to be run prior to giving the presentation.

The culmination of interactive analysis is a bokeh dashboard, which is a web interface outside of the jupyter notebook. The command to run this is in the final cell of the notebook, commented out since it never stops running.

sdanalysis figure -m model/knn-trimer.pkl --directory data/dataset