Command Line Interface

The AlphaPy Pro Command Line Interface (CLI) provides a streamlined way to run machine learning pipelines. Simply navigate to your project directory and run alphapy.

Basic Usage

First, change to your project directory:

cd path/to/project

Train models and generate predictions:

alphapy

Core Commands

alphapy - Main ML Pipeline

Usage:

alphapy [options]

Options:

--train - Train new models and make predictions [Default]
--predict - Make predictions using saved models
--verbose - Enable verbose logging
--debug - Enable debug mode with detailed output

Example:

# Train models (default behavior)
alphapy

# Use a saved model for predictions
alphapy --predict

# Train with verbose output
alphapy --train --verbose

Custom Pipelines

You can create custom data-preparation pipelines around AlphaPy by producing the canonical train/test inputs expected by the package and then running alphapy from the project directory.

Project Structure Requirements

Before running commands, ensure your project follows this structure:

my_project/
├── config/
│   └── model.yml       # Model configuration
├── data/
│   ├── train.csv       # Training data
│   └── test.csv        # Testing data (optional)
└── runs/               # Output directory (auto-created)

Output Structure

After running a command, outputs are organized in timestamped directories:

runs/
└── run_YYYYMMDD_HHMMSS/
    ├── config/         # Configuration used
    ├── input/          # Input data snapshots
    ├── model/          # Trained models and metrics
    ├── output/         # Predictions and rankings
    └── plots/          # Visualizations

Common Workflows

1. Quick Model Training:

cd projects/my_project
alphapy

2. Hyperparameter Tuning:

First, edit config/model.yml to enable grid search:

model:
    grid_search:
        option: True
        iterations: 100

Then run:

alphapy --verbose

3. Batch Predictions:

# Train once
alphapy --train

# Use the model for multiple prediction sets
alphapy --predict

Logging and Debugging

AlphaPy Pro creates detailed logs:

alphapy.log - Main pipeline log

To increase log verbosity:

alphapy --verbose --debug

Tips and Best Practices

Always verify your data before running models
Start with small iterations for grid search, then increase
Use dated runs for reproducibility
Keep model.yml under version control
Review plots in the output directory for insights

For more details on configuration options, see Project Structure.