Skip to content

CLAUDE.md

Latest commit

 

History

History
77 lines (56 loc) · 3.03 KB

CLAUDE.md

File metadata and controls

77 lines (56 loc) · 3.03 KB
Raw

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Environment

  • Python 3.14.3 in a virtual environment named myenv on macOS (Apple Silicon M5)
  • Primary IDE: JupyterLab — almost all work lives in .ipynb notebooks
  • The active environment is activated with source myenv/bin/activate

Setup

# Create and activate the virtual environment
python3 -m venv myenv
source myenv/bin/activate

# Core dependencies
pip install pandas numpy matplotlib seaborn scikit-learn statsmodels pmdarima scipy
pip install openai anthropic jupyterlab yfinance sqlalchemy

# Launch JupyterLab
jupyter lab

API keys are loaded from the shell environment, never hardcoded:

export ANTHROPIC_API_KEY='sk-ant-...'   # add to ~/.zshrc

Repository Structure

Folder Contents
python_vignettes/ Language fundamentals, pandas, numpy, plotting, stats, OOP, R-to-Python (_r2p suffix) conversions
machine_learning/ Supervised learning notebooks following ISLR chapters (Ch3, Ch4, Ch8, Ch10); Kaggle examples; LLM hello world; agentic AI vignette
toolbox/ Reusable standalone notebooks: descriptive stats, prediction metrics, visualization helpers
use_cases/ End-to-end applied analyses on public datasets (migration, FIFA, markets)
data/ CSV/XLS/SQLite datasets used across notebooks

Notebook Conventions

Every notebook opens with a module-level docstring (Cell [1]) that includes:

  • @author, @institution, @environment, description, and revision history
  • A structured Revision History: block with dates and bullet points

.py companion files

Some notebooks have a paired .py file (e.g. python_functions_vignette.py alongside .ipynb). The python_vignettes/ folder also has importable modules:

  • my_stats_ftns_module.py — hand-rolled mean_s() / stdev_s() using only Python built-ins
  • functions_vignette_library.py — function-pattern library imported by python_functions_vignette.ipynb
  • my_stats_ftns_calls.py — caller script for the stats module

Naming Conventions

  • _r2p suffix — notebooks that are direct R-to-Python translations
  • _vignette suffix — focused demonstrations of a single concept or library
  • _oo suffix — object-oriented implementations
  • _ISLR_ChN — tied to a specific chapter of Introduction to Statistical Learning with Python

Agentic AI Pattern (machine_learning/agentic_ai_vignette_yfinance.ipynb)

The agent loop pattern used here:

  1. Define Python tool functions (get_stock_price, get_portfolio_summary backed by yfinance)
  2. Register JSON tool schemas with Claude
  3. Loop: send message → receive tool_use → execute Python function → feed result back → repeat until end_turn

Uses claude-haiku-4-5-20251001 for speed/cost. The dispatch table (TOOL_FUNCTIONS) maps tool names to callables.

Git Workflow

git add <file>
git commit -m "Descriptive message"
git push

See git_cheatsheet.md for PAT renewal and full reference.