Plausibilities

Fundamentals

Upcoming

• mybinder.org
• CodeSpaces
• GitHub Actions
• Docker

Analysing, modelling, forecasting, etc., amidst uncertainty.

Content

• Studies
• Development Notes
  • Environment
  • Environment Updates
  • Requirements
  • Graphing
• References

Studies

Notebooks:

• Regression

  • A contrasts of linear regression models
    

  • Simple linear regression
    

  • Robust linear regression
    

• Gaussian Processes

  • Gaussian process regression
    

Development Notes

Environment

The environment of the projects of Uncertainty

conda create --prefix ~/Anaconda3/envs/uncertainty

The installations are

conda install -c anaconda pymc3 # installs: python, theano, arviz, numpy, pandas
conda install -c anaconda seaborn # installs: matplotlib
conda install -c anaconda python-graphviz # installs: graphviz
conda install -c anaconda pywin32 jupyterlab nodejs # installs: requests, urllib3
pip install dotmap


# For norms & testing
conda install -c anaconda pytest coverage pylint pytest-cov

# Upgrading PyMC3
pip install --upgrade pymc3==3.9.3

Environment Updates

Update python via conda install -c anaconda python==3.8.13, numpy via conda update -c anaconda numpy, pandas via conda update -c anaconda pandas; for help visit conda commands.   In terms of pymc3 & pymc ↠ pip unistall pymc3, subsequently pip install pymc.   Update pymc via pip install --upgrade pymc.   An alternative [general] update option, within an environment, is

  conda update -c anaconda --update-all

Extra

  conda install -c anaconda colorcet
  conda install -c anaconda m2w64-toolchain_win-64

Requirements

For project fundamentals

conda activate uncertainty
pip freeze -r docs/filter.txt > requirements.txt

and

pylint --generate-rcfile > .pylintrc

Graphing

Aesthetics

• Matplotlib Style Sheets, e.g., plt.style.use('fivethirtyeight')

• Arviz, e.g., az.style.use('arviz-darkgrid')

• Seaborn, e.g., sns.set_style("darkgrid"), sns.set_context("poster"), sns.set_color_codes("pastel")

Layouts

• Constrained Layout

• matplotlib.rcParams['figure.constrained_layout.use'] = False

• plt.rcParams['figure.constrained_layout.use'] = False

• plt.figure(constrained_layout=False)

• Adjust: plt.tight_layout(pad=1.0, h_pad=1.5, w_pad=1.5, rect=(0,0,1,1))

References

• PyMC
• actions/checkout@v2
• GitHub Glossary
• Google Colaboratory & GitHub
• Seaborn Graph Types
• pip commands
• conda commands
• The Anaconda repository