How-to Guides

Short, task-oriented guides for common pyjpt workflows.

• How to Work with Variables
  • Variables and Domains
    • Domain Factory Functions
  • Variable Types
    • NumericVariable
    • IntegerVariable
    • SymbolicVariable
  • Inferring Variables from a DataFrame
  • Labels vs. Values
  • Variable Maps and Assignments
    • VariableMap
    • VariableAssignment
  • Impurity Inversion for Symbolic Variables
    • When to Use Impurity Inversion
    • Example
  • Variable Settings
  • Serialization
• How to Classify with JPTs
  • Problem Setup
  • Training a Discriminative JPT
  • Making Predictions
  • Evaluating Accuracy
• How to Predict Continuous Values with JPTs
  • Problem Setup
  • Training a Discriminative JPT
  • Point Predictions via Expectation
  • Full Posterior Distribution
  • Evaluating RMSE
• How to Use Dependency Discovery and Xi Pruning
  • Mathematical Background
    • Chatterjee’s \(\xi\) coefficient
    • Significance test
  • Dependency Discovery
    • Basic usage
    • Persistence
    • Backward compatibility
  • Xi-Based Pruning
    • Using the pruning criterion
  • Combining Both
  • Worked Example
  • Extending with Custom Discovery Strategies
  • References
• How to Use Split Validation
  • Basic Usage
  • Choosing a Mode
  • min_eval_samples — Require a Minimum of Held-out Rows per Child
  • Serialisation
  • Troubleshooting
• How to Visualise a JPT
  • Plotting the Tree Structure
  • Plotting Individual Distributions
  • Saving Figures
  • Custom Rendering Engine
• How to Save and Load a JPT
  • Saving with Pickle (default)
  • Loading with Pickle
  • Saving and Loading with JSON
  • Serialising to a Bytes Buffer
  • Choosing a Format