Examples

Customer Churn Classification

from universal_ml_framework import UniversalMLPipeline, DataGenerator

# Generate sample data
DataGenerator.generate_customer_churn()

# Run classification pipeline
pipeline = UniversalMLPipeline(problem_type='classification')
pipeline.run_pipeline(
    train_path='data/customer_train.csv',
    target_column='Churn',
    test_path='data/customer_test.csv'
)

print(f"Best model: {pipeline.best_model_name}")
print(f"Best score: {pipeline.best_score:.4f}")

House Price Prediction

from universal_ml_framework import UniversalMLPipeline, DataGenerator

# Generate sample data
DataGenerator.generate_house_prices()

# Run regression pipeline
pipeline = UniversalMLPipeline(problem_type='regression')
pipeline.run_pipeline(
    train_path='data/house_train.csv',
    target_column='SalePrice',
    test_path='data/house_test.csv'
)

print(f"Best model: {pipeline.best_model_name}")
print(f"Best MSE: {pipeline.best_score:.2f}")

Sales Forecasting

from universal_ml_framework import UniversalMLPipeline, DataGenerator

# Generate sample data
DataGenerator.generate_sales_forecasting()

# Run regression pipeline
pipeline = UniversalMLPipeline(problem_type='regression')
pipeline.run_pipeline(
    train_path='data/sales_train.csv',
    target_column='Sales',
    test_path='data/sales_test.csv'
)

Using Predefined Configurations

from universal_ml_framework import run_pipeline_with_config, list_available_configs

# List available configurations
list_available_configs()

# Run with predefined config
pipeline = run_pipeline_with_config('customer_churn')

Custom Feature Engineering

from universal_ml_framework import UniversalMLPipeline

pipeline = UniversalMLPipeline(problem_type='classification')

# Manually specify feature types
pipeline.feature_types = {
    'numeric': ['age', 'income', 'tenure'],
    'categorical': ['city', 'job_type', 'education'],
    'binary': ['has_phone', 'has_internet', 'is_senior']
}

pipeline.run_pipeline(
    train_path='data.csv',
    target_column='target',
    custom_features=pipeline.feature_types['numeric'] +
                   pipeline.feature_types['categorical'] +
                   pipeline.feature_types['binary']
)

Batch Processing Multiple Datasets

from universal_ml_framework import UniversalMLPipeline, DataGenerator

# Generate all sample datasets
DataGenerator.generate_all_datasets()

datasets = [
    ('data/customer_train.csv', 'Churn', 'classification'),
    ('data/house_train.csv', 'SalePrice', 'regression'),
    ('data/sales_train.csv', 'Sales', 'regression')
]

results = {}
for train_path, target, problem_type in datasets:
    pipeline = UniversalMLPipeline(problem_type=problem_type)
    pipeline.run_pipeline(train_path, target)
    results[train_path] = {
        'best_model': pipeline.best_model_name,
        'best_score': getattr(pipeline, 'best_score', 'N/A')
    }

for dataset, result in results.items():
    print(f"{dataset}: {result['best_model']} - {result['best_score']}")

Working with Your Own Data

from universal_ml_framework import UniversalMLPipeline

# For your own CSV file
pipeline = UniversalMLPipeline(problem_type='classification')
pipeline.run_pipeline(
    train_path='your_data.csv',
    target_column='your_target_column',
    test_path='your_test_data.csv',
    exclude_columns=['id', 'timestamp', 'irrelevant_column']
)

# Check results
print(f"Best model: {pipeline.best_model_name}")
print(f"Cross-validation score: {pipeline.cv_results[pipeline.best_model_name]['mean']:.4f}")
print(f"Feature types detected: {pipeline.feature_types}")

Loading Saved Models

import joblib
import json

# Load saved model
model = joblib.load('best_model.pkl')

# Load model metadata
with open('model_info.json', 'r') as f:
    model_info = json.load(f)

print(f"Model type: {model_info['best_model']}")
print(f"Problem type: {model_info['problem_type']}")
print(f"CV Score: {model_info['cv_score']:.4f}")

# Make predictions on new data
# predictions = model.predict(new_data)

Tip

All examples generate output files that you can examine:

• predictions.csv - Model predictions

• best_model.pkl - Trained model

• model_info.json - Model metadata and performance