Example Notebooks¶

Pollywog includes a comprehensive collection of Jupyter notebooks demonstrating common workflows and advanced features. All examples are available in the examples/ folder on GitHub and can be run interactively in your browser at JupyterLite.

Note

Try examples in your browser!

Visit https://endarthur.github.io/pollyweb to run all examples interactively without installing anything. Perfect for learning and experimentation!

Getting Started¶

For beginners, start with these notebooks in order:

Basic Usage - Core concepts and file I/O
Item Types Guide - Understanding Variable, Number, Category, Filter
Helper Functions Showcase - Using helper functions for common patterns

For experienced users, explore specific topics:

Conditional Logic - Advanced If statements and nested conditions
Complete Mining Workflow - Complete mining workflow example
Dependency Analysis - Understanding calculation dependencies
Modifying .lfcalc Files - Read, modify, and save existing .lfcalc files

Core Concepts¶

Basic Usage¶

File: basic_usage.ipynb

Topics covered:

Creating CalcSet objects
Creating Number, Category, and Filter calculations
Writing to .lfcalc files
Reading existing .lfcalc files
Postprocessing variable estimates
Domain-weighted calculations
Thresholding and classification
Scikit-learn model conversion

Best for: First-time pollywog users, general reference

from pollywog.core import CalcSet, Number

calcset = CalcSet([
    Number("Au_diluted", "[Au_est] * 0.95")
])
calcset.to_lfcalc("output.lfcalc")

Item Types Guide¶

File: item_types_guide.ipynb

Topics covered:

Variable vs Number vs Category vs Filter
When to use each type
Visibility in Leapfrog (calculator vs everywhere)
Decision flow for choosing types
Complete workflow using all types

Best for: Understanding which item type to use when

Type	Visible Outside Calculator?	Use Case
Variable	❌ No	Intermediate calculations
Number	✅ Yes	Final numeric outputs
Category	✅ Yes	Classifications and labels
Filter	✅ Yes	Boolean inclusion rules

Conditional Logic¶

File: conditional_logic.ipynb

Topics covered:

Basic If/otherwise patterns
Multiple conditions
Complex conditions with logical operators
Nested If statements
Mill feed decision trees
Handling missing data
Practical mining examples

Best for: Building complex conditional logic

from pollywog.core import Number, Category, If

# Multi-class classification
ore_class = Category(
    "ore_class",
    If(
        [
            ("[Au] < 0.3", "'waste'"),
            ("[Au] < 1.0", "'low_grade'"),
            ("[Au] < 3.0", "'medium_grade'"),
        ],
        "'high_grade'"
    )
)

Helper Functions¶

Helper Functions Showcase¶

File: helper_functions.ipynb

Topics covered:

Mathematical helpers (Sum, Product, Average)
WeightedAverage (critical for multi-domain modeling!)
WeightedSum
Scale (dilution, recovery, unit conversion)
Normalize (proportions)
CategoryFromThresholds (grade classification)
Dual-mode usage (with/without name parameter)
Composing helpers for complex expressions

Best for: Learning all available helper functions

from pollywog.helpers import WeightedAverage, Scale, CategoryFromThresholds

# Domain-weighted grade
au_composite = WeightedAverage(
    variables=["Au_oxide", "Au_sulfide", "Au_transition"],
    weights=["prop_oxide", "prop_sulfide", "prop_transition"],
    name="Au_composite"
)

# Apply dilution
au_diluted = Scale("Au_composite", 0.95, name="Au_diluted")

# Classify
ore_class = CategoryFromThresholds(
    variable="Au_diluted",
    thresholds=[0.5, 2.0],
    categories=["waste", "low_grade", "high_grade"],
    name="ore_class"
)

Complete Workflows¶

Complete Mining Workflow¶

File: pollywog_workflow_tutorial.ipynb

Topics covered:

Preprocessing drillhole data
Postprocessing block model grades
Domain-weighted calculations
Scikit-learn regression trees for recovery
Helper functions in practice
Visualization in Jupyter

Best for: End-to-end resource estimation workflow

Dependency Analysis¶

File: dependency_analysis.ipynb

Topics covered:

Automatic dependency extraction
Topological sorting (resolving calculation order)
External dependencies (variables from Leapfrog)
Circular dependency detection
Analyzing complex workflows
Debugging dependency issues

Best for: Understanding and debugging calculation order

# Dependencies are automatically extracted
calc = Number("result", "[a] + [b] * [c]")
print(calc.dependencies)  # {'a', 'b', 'c'}

# Sort calculations by dependencies
sorted_calcset = calcset.topological_sort()

Modifying .lfcalc Files¶

File: modifying_lfcalc_files.ipynb

Topics covered:

Reading .lfcalc files
Querying and filtering calculations
Modifying existing calculations
Adding new calculations
Removing calculations
Batch modifications
Merging multiple files
Version control workflows

Best for: Working with existing Leapfrog calculations

# Load existing file
calcset = CalcSet.read_lfcalc("existing.lfcalc")

# Query by name
au_calcs = calcset.query('name.startswith("Au")')

# Modify
for item in calcset.items:
    if "diluted" in item.name:
        item.expression = ["[Au_est] * 0.95"]

# Save
calcset.to_lfcalc("modified.lfcalc")

Advanced Topics¶

Querying CalcSets¶

File: querying_calcsets.ipynb

Topics covered:

Query method basics
Filtering by name, type, or attributes
Using external variables in queries
Practical query examples

# Query by name pattern
subset = calcset.query('name.startswith("Au")')

# Query by comment
estimates = calcset.query('"estimate" in comment_equation')

# Use external variable
prefix = "Cu"
subset = calcset.query('name.startswith(@prefix)')

Running with DataFrames¶

File: running_with_dataframe.ipynb

Topics covered:

Running calculations on pandas DataFrames
Using run_calcset() function
Using pandas accessor (df.pw.run())
Testing calculations before deploying

import pandas as pd
from pollywog.run import run_calcset

# Test on DataFrame
df = pd.DataFrame({"Au": [1.0, 2.0, 3.0]})
result = run_calcset(calcset, dataframe=df)

# Or use accessor
result = df.pw.run(calcset)

Scikit-learn Conversion¶

File: sklearn_conversion.ipynb

Topics covered:

Converting decision trees to calculations
Converting random forests
Converting linear models
Inspecting converted calculations

from sklearn.tree import DecisionTreeRegressor
from pollywog.conversion.sklearn import convert_tree

# Train model
reg = DecisionTreeRegressor().fit(X, y)

# Convert to calculation
calc = convert_tree(reg, feature_names, "prediction")

# Add to CalcSet
calcset = CalcSet([calc])

Display and Visualization¶

File: display_final_workflow.ipynb

Topics covered:

Loading and displaying .lfcalc files
Interactive visualization in Jupyter
Theming (light/dark mode)

from pollywog.display import display_calcset, set_theme

set_theme("dark")
display_calcset(calcset)

JupyterLite Specific¶

JupyterLite Quickstart¶

File: jupyterlite_quickstart.ipynb

Topics covered:

Running pollywog in the browser
Creating simple workflows
Exporting files for download

Best for: First-time JupyterLite users

JupyterLite Demo¶

File: jupyterlite_demo.ipynb

Topics covered:

Pollywog magic commands
Autodownload feature
Manual download utilities
Working with browser storage

# Enable autodownload
%load_ext pollywog.magics
%pollywog autodownload on

# Now exports trigger download
calcset.to_lfcalc("output.lfcalc")

Quick Reference by Task¶

I want to…

Task	See Example
Create my first calculation set	Basic Usage
Understand Variable vs Number vs Category	Item Types Guide
Use If statements and conditions	Conditional Logic
Apply dilution and recovery factors	Helper Functions Showcase (Scale helper)
Combine domain estimates	Helper Functions Showcase (WeightedAverage)
Classify by grade thresholds	Helper Functions Showcase (CategoryFromThresholds)
Build a complete resource workflow	Complete Mining Workflow
Fix calculation order issues	Dependency Analysis
Update existing .lfcalc files	Modifying .lfcalc Files
Filter calculations by name	Querying CalcSets
Test calculations on data	Running with DataFrames
Use machine learning models	Scikit-learn Conversion
Try pollywog without installing	JupyterLite Quickstart

Running the Examples¶

Option 1: JupyterLite (No Installation)

Visit https://endarthur.github.io/pollyweb and open any notebook. Everything runs in your browser!

Option 2: Local Installation

# Install pollywog
pip install lf_pollywog

# Clone repository for examples
git clone https://github.com/endarthur/pollywog.git
cd pollywog/examples

# Start Jupyter
jupyter notebook

Option 3: Individual Files

Download individual notebooks from GitHub and run locally.

Next Steps¶

After exploring the examples:

Read the guides:
- Leapfrog Expression Syntax Guide - Leapfrog expression syntax reference
- Helper Functions Guide - Detailed helper function documentation
- Common Workflow Patterns - Common patterns and best practices
Check the API:
- API Reference - Complete API documentation
Get help:
- GitHub Issues - Report bugs or request features
- GitHub Discussions - Ask questions and share ideas

Example Notebooks¶

Getting Started¶

Core Concepts¶

Basic Usage¶

Item Types Guide¶

Conditional Logic¶

Helper Functions¶

Helper Functions Showcase¶

Complete Workflows¶

Complete Mining Workflow¶

Dependency Analysis¶

Modifying .lfcalc Files¶

Advanced Topics¶

Querying CalcSets¶

Running with DataFrames¶

Scikit-learn Conversion¶

Display and Visualization¶

JupyterLite Specific¶

JupyterLite Quickstart¶

JupyterLite Demo¶

Quick Reference by Task¶

Running the Examples¶

Next Steps¶

pollywog

Navigation

Related Topics