Quick Start Guide

Basic Usage

Classification Problem

from universal_ml_framework import UniversalMLPipeline

# Create pipeline
pipeline = UniversalMLPipeline(problem_type='classification')

# Run complete pipeline
pipeline.run_pipeline(
    train_path='train.csv',
    target_column='target',
    test_path='test.csv'
)

Regression Problem

from universal_ml_framework import UniversalMLPipeline

# Create pipeline
pipeline = UniversalMLPipeline(problem_type='regression')

# Run complete pipeline
pipeline.run_pipeline(
    train_path='train.csv',
    target_column='price',
    test_path='test.csv'
)

Quick Setup Functions

One-liner Classification

from universal_ml_framework import quick_classification_pipeline

result = quick_classification_pipeline('data.csv', 'target_column')

One-liner Regression

from universal_ml_framework import quick_regression_pipeline

result = quick_regression_pipeline('data.csv', 'price_column')

Generate Sample Data

from universal_ml_framework import DataGenerator

# Generate synthetic datasets
DataGenerator.generate_customer_churn()
DataGenerator.generate_house_prices()
DataGenerator.generate_sales_forecasting()

# Generate all datasets at once
DataGenerator.generate_all_datasets()

Customization Options

Exclude Columns

pipeline.run_pipeline(
    train_path='data.csv',
    target_column='target',
    exclude_columns=['id', 'timestamp', 'name']
)

Custom Feature Types

pipeline.feature_types = {
    'numeric': ['age', 'income', 'score'],
    'categorical': ['city', 'category', 'type'],
    'binary': ['has_feature', 'is_active']
}

What Happens Automatically

Note

The framework handles the entire ML pipeline automatically:

Data Loading - Reads CSV files
Feature Detection - Identifies feature types
Preprocessing - Handles missing values, encoding, scaling
Model Training - Tests multiple algorithms
Cross Validation - Evaluates model performance
Hyperparameter Tuning - Optimizes best model
Prediction - Generates test predictions
Model Saving - Persists trained model

Tip

Check the generated files after running:

predictions.csv - Test predictions
best_model.pkl - Trained model
model_info.json - Model metadata