Claude.md for Python Data Science

# Project: Customer Analytics Pipeline


## Development Standards

- **Language**: Python 3.11+

- **Code Style**: Follow PEP 8 strictly, use Black for formatting

- **Type Hints**: Required for all function signatures and class definitions

- **Documentation**: Docstrings required for all public functions and classes


## Workflow Requirements

1. Create feature branch: `analysis-[description]` or `model-[description]`

2. Write unit tests for all data processing functions

3. Run `pytest` and ensure all tests pass

4. Run `black .` and `flake8` before committing

5. Update relevant documentation in `/docs` if adding new features


## Project Structure

- `/src/data`: Data ingestion and preprocessing modules

- `/src/models`: ML model definitions and training scripts  

- `/src/analysis`: Exploratory analysis notebooks and scripts

- `/src/utils`: Shared utility functions

- `/tests`: Comprehensive test suite

- `/configs`: Configuration files for different environments


## Data Handling Standards

- Use Pandas for data manipulation, prefer vectorized operations

- All data files must be documented in `/data/README.md`

- Use Pydantic models for data validation and serialization

- Never commit raw data files to version control

- Use environment variables for database connections and API keys


## ML/Analysis Guidelines

- Use scikit-learn for standard ML algorithms

- Notebook naming: `YYYY-MM-DD-[initials]-[description].ipynb`

- Save all trained models with versioning in `/models/trained`

- Use MLflow for experiment tracking

- Include model performance metrics in commit messages


## Dependencies

- Core: pandas, numpy, scikit-learn, matplotlib, seaborn

- ML: xgboost, lightgbm, optuna

- Data: sqlalchemy, pydantic, requests

- Testing: pytest, pytest-cov

Comments

Post a Comment

Popular posts from this blog

Worktree Prompt (Separate Directories for Different Features - So You Can Work in Parallel)

NOTE : click on the code sections to copy ( Save this as .claude/commands/parallel-work.md) I want to develop features in parallel for my expense tracker app using Git worktrees: $ARGUMENTS Think about how to divide the work up into separate features unless this has been clearly explained already. Please help me set up the worktree environment: 1. For each feature mentioned, create a worktree at ../expense-tracker-[feature-name] with branch feature/[feature-name] 2. Set up the development environment in each worktree 3. List all worktrees to confirm they were created 4. Explain what each worktree will contain and how they're isolated I want to be able to work on all features simultaneously without conflicts. That's for creation. When you're done: (Save this as .claude/commands/integrate-parallel-work.md) I have features developed in parallel worktrees that I need to integrate: $ARGUM...
Read more

Claude.md for Responsive Mobile App Development with React Native

 # Project: FitnessTracker Mobile App ## Platform Requirements - **Framework**: React Native 0.72+ with TypeScript - **State Management**: Redux Toolkit with RTK Query - **Navigation**: React Navigation v6 - **UI Library**: Native Base for consistent design - **Platform Support**: iOS 14+ and Android API 24+ ## Branch Strategy 1. Create feature branches: `mobile-[feature-name]` 2. Use conventional commits: `feat:`, `fix:`, `refactor:`, etc. 3. Test on both iOS and Android before committing 4. Run `npm run lint` and `npm run type-check` before commits ## Code Organization - **Screens**: `/src/screens/[FeatureName]/[ScreenName]Screen.tsx` - **Components**: `/src/components/[ComponentName]/index.tsx` - **Navigation**: `/src/navigation/[NavigatorName]Navigator.tsx` - **Services**: `/src/services/[serviceName].ts` - **Store**: `/src/store/slices/[featureName]Slice.ts` ## Mobile-Specific Standards - Use React Native's built-in components before third-party libraries - Implement proper er...
Read more