Price Prediction ML Pipeline

Quick Start

1. Install dependencies:

pip install uv
uv pip install -r requirements.txt

2. Initialize you ZenML repo

zenml init
zenml login
zenml project set <INSERT_NAME>
zenml stack set <INSERT_NAME>

2. Run training pipeline:

# Using default configuration
python training_pipeline.py

# Using scheduled configuration (runs every minute for testing)
python training_pipeline.py --config training_schedule_config.yaml

# Using custom configuration
python training_pipeline.py --config path/to/your/config.yaml

3. Run inference:

python inference_pipeline.py

Schedule Management

Deploy Scheduled Pipeline

# Create and deploy a scheduled training pipeline (runs every minute for testing)
python training_pipeline.py --config training_schedule_config.yaml

Manage Schedules

# List all schedules
zenml pipeline schedule list
# Update schedule cron expression
zenml pipeline schedule update price-prediction-training-schedule --cron-expression="*/2 * * * *"

# Delete a schedule (removes from both ZenML and orchestrator)
zenml pipeline schedule delete price-prediction-training-schedule

Current Implementation

This is a proof-of-concept ML pipeline using ZenML for price prediction with synthetic e-commerce data.

Potential Improvements

• Extract pipeline configurations from decorators to YAML files
• Move pipeline steps into individual module files
• Use Model stages to attach the inference pipeline only to the
• Connect to real data sources instead of synthetic data
• Add proper configuration management and environment variables
• Implement comprehensive testing and validation
• Add proper logging and monitoring
• Separate business logic from infrastructure concerns
• Add input validation and error handling
• Implement proper CI/CD pipeline