Diagrams

All Hail IKEA

Billy Bookcase Assembly Instructions

Bayesian Modeling Workflow

A workflow diagram for bayesian modeling. Step 1: setting up a probability model, with model and prior specifications. Prior predictive checking. This may result in sensitivity/identifiability preliminary analyses before continuing to step 2. Step 2: Learning. Calibration and convergence diagnostics.  Step 3: Model checking and validation, including residual analysis and posterior predictive model checks. This leads to estimation of prediction accuracy and interpretation of parameter values.

Image source: Building energy statistical modeling ch. 3

Scikit Learn estimator workflow

A map of data characteristics (>50 samples, < 10k samples, <100k samples, categorical, continuous, labeled, text) and goals (predicting a category/quantity, predicting structure) that leads to four basic groups of techniques: classification, clustering, dimension reduction, and regression.

Image source: Scikit Learn user guide

Useful Tools

  • Excalidraw (online)

  • Draw.io (online)

  • Inkscape (Free/open source)