🧠 CANS: Causal Adaptive Neural System

Production-ready causal inference at scale with deep learning, APIs, and LLM integration

CANS (Causal Adaptive Neural System) is the most comprehensive production-ready framework for causal inference using deep learning. It combines Graph Neural Networks (GNNs), Transformers, Counterfactual Regression Networks (CFRNet), and advanced causal methods with enterprise-grade APIs and LLM integration.

🎯 Perfect for: Healthcare, Finance, Marketing, Legal, Social Media, E-commerce, and any domain requiring rigorous causal analysis

🌟 What Makes CANS Unique

• 🧠 Hybrid AI Architecture: GNNs + Transformers + CFRNet for complex data
• 🔬 Rigorous Causal Science: Automated assumption testing, multiple identification strategies
• 🌐 Production APIs: REST API + MCP server for seamless integration
• 🤖 LLM Integration: Enable AI assistants to perform causal analysis autonomously
• ⚡ Enterprise Ready: Authentication, monitoring, scalable deployment
• 📊 Comprehensive Toolkit: CLI, Python API, web integration, notebooks

🚀 Choose Your Interface

Interface	Best For	Getting Started
📚 Interactive Tutorials	Learning, first-time users	cans-tutorials → hands-on guided learning
🖥️ CLI Tools	Quick analysis, data scientists	cans-validate --data data.csv --treatment T --outcome Y
🐍 Python API	Research, notebooks, pipelines	from cans import CANS, validate_causal_assumptions
🌐 REST API	Web apps, microservices	POST /validate → JSON response
🤖 MCP Server	LLMs, AI assistants	Claude/GPT calls validate_causal_assumptions_tool

📊 Usage Matrix

Task	CLI	Python	REST API	MCP/LLM
Assumption Validation	cans-validate	validate_causal_assumptions()	POST /validate	validate_causal_assumptions_tool
Complete Analysis	cans-analyze	CANSRunner.fit()	POST /analyze	quick_causal_analysis
Model Evaluation	cans-evaluate	CausalEvaluator.evaluate()	POST /evaluate	evaluate_predictions
Batch Processing	Shell scripting	for loops	HTTP requests	LLM automation
Production Deployment	Cron jobs	Python services	Kubernetes	AI workflows

🚀 What's New in v3.0 - Production-Ready Causal AI

🔬 Advanced Causal Methods

• Assumption Testing: Automated unconfoundedness, positivity, and SUTVA validation
• Multiple Identification: Backdoor criterion, IPW, doubly robust estimation
• CATE Estimation: X-Learner, T-Learner, S-Learner, Neural CATE, Causal Forest
• Uncertainty Quantification: Bayesian methods, ensemble approaches, conformal prediction

🌐 Enterprise Integration

• REST API: Complete HTTP API with authentication and rate limiting
• MCP Server: Model Context Protocol for seamless LLM integration
• Production Tools: Docker, Kubernetes, monitoring, logging
• Multi-language: Python, JavaScript, R, cURL examples

🧠 Enhanced AI Architecture

• Advanced Graph Construction: Multi-node, temporal, and global architectures
• Causal-Specific Losses: CFR, IPW, DragonNet, TARNet with representation balancing
• Memory Efficiency: Lazy loading and batch processing for large datasets
• GPU Optimization: CUDA support with automatic device selection

Previous v2.0 Features:

• 🔧 Configuration Management: Centralized, validated configs with JSON/YAML support
• 🛡️ Enhanced Error Handling: Comprehensive validation with informative error messages
• 📊 Logging & Checkpointing: Built-in experiment tracking with automatic model saving
• 🧪 Comprehensive Testing: 100+ unit tests ensuring production reliability
• 📈 Advanced Data Pipeline: Multi-format loading (CSV, JSON) with automatic preprocessing
• ⚡ Enhanced Training: Early stopping, gradient clipping, multiple loss functions

🔧 Key Features

Core Architecture

• ✅ Hybrid Neural Architecture: GNNs + Transformers + CFRNet for multi-modal causal inference
• ✅ Gated Fusion Layer: Adaptive mixing of graph and textual representations
• ✅ Flexible Graph Construction: Single-node, multi-node, temporal, and global graphs
• ✅ Production-Ready: Comprehensive error handling, logging, and testing

Causal Inference Capabilities

• ✅ Rigorous Assumption Testing: Automated validation of causal identification conditions
• ✅ Multiple Identification Methods: Backdoor, IPW, doubly robust, with sensitivity analysis
• ✅ Heterogeneous Treatment Effects: CATE estimation with 5+ methods (X/T/S-Learners, etc.)
• ✅ Advanced Loss Functions: CFR, DragonNet, TARNet with representation balancing
• ✅ Uncertainty Quantification: Bayesian, ensemble, conformal prediction approaches

Data Processing & Evaluation

• ✅ Smart Data Loading: CSV, JSON, synthetic data with automatic graph construction
• ✅ Comprehensive Evaluation: PEHE, ATE, policy value, calibration metrics
• ✅ Memory Efficiency: Lazy loading, batch processing for large-scale datasets
• ✅ Easy Configuration: JSON/YAML configs with validation and experiment tracking

🏗️ Architecture

 +-----------+     +-----------+
 |  GNN Emb  |     |  BERT Emb |
 +-----------+     +-----------+
        \             /
         \ Fusion Layer /
          \     /
         +-----------+
         |  Fused Rep |
         +-----------+
               |
           CFRNet
        /          \
   mu_0(x)       mu_1(x)

🚀 Enhanced Causal Analysis Workflow

Complete Example with Assumption Testing & CATE Estimation

from cans import (
    CANSConfig, CANS, GCN, CANSRunner,
    create_sample_dataset, get_data_loaders,
    CausalAssumptionTester, CausalLossManager, 
    CATEManager, UncertaintyManager,
    advanced_counterfactual_analysis
)

# 1. Configuration with enhanced causal features
config = CANSConfig()
config.model.gnn_type = "GCN"
config.training.loss_type = "cfr"  # Causal loss function
config.data.graph_construction = "global"  # Multi-node graphs

# 2. Test causal assumptions BEFORE modeling
assumption_tester = CausalAssumptionTester()
results = assumption_tester.comprehensive_test(X, T, Y)
print(f"Causal assumptions valid: {results['causal_identification_valid']}")

# 3. Create datasets with enhanced graph construction
datasets = create_sample_dataset(n_samples=1000, config=config.data)
train_loader, val_loader, test_loader = get_data_loaders(datasets)

# 4. Setup model with causal loss functions
from transformers import BertModel
gnn = GCN(in_dim=64, hidden_dim=128, output_dim=256)
bert = BertModel.from_pretrained("distilbert-base-uncased")
model = CANS(gnn, bert, fusion_dim=256)

loss_manager = CausalLossManager("cfr", alpha=1.0, beta=0.5)

# 5. Train with causal-aware pipeline
optimizer = torch.optim.AdamW(model.parameters(), lr=0.001)
runner = CANSRunner(model, optimizer, config)
history = runner.fit(train_loader, val_loader)

# 6. Multiple counterfactual identification methods
cf_results = advanced_counterfactual_analysis(
    model, test_loader, 
    methods=['backdoor', 'ipw', 'doubly_robust']
)

# 7. CATE estimation with multiple learners
cate_manager = CATEManager(method="x_learner")
cate_manager.fit(X, T, Y)
individual_effects = cate_manager.estimate_cate(X_test)

# 8. Uncertainty quantification
uncertainty_manager = UncertaintyManager(method="conformal")
uncertainty_manager.setup(model)
intervals = uncertainty_manager.estimate_uncertainty(test_loader)

print(f"ATE: {cf_results['backdoor']['ate']:.3f}")
print(f"Coverage: {intervals['coverage_rate']:.3f}")

🚀 Quick Start

🎯 New to CANS? Start with our Getting Started Guide for a 5-minute tutorial!

Installation

# Install from PyPI (Recommended)  
pip install cans-framework

# Verify installation and set up tutorials
pip show cans-framework
cans-tutorials  # Sets up interactive learning environment

Alternative Installation Methods

# Development installation
git clone https://github.com/rdmurugan/cans-framework.git
cd cans-framework
pip install -r requirements.txt
pip install -e .

# With conda (for dependency management)
conda create -n cans python=3.9
conda activate cans
pip install cans-framework

Core Dependencies:

• torch>=2.0.0
• transformers>=4.38.0
• torch-geometric>=2.3.0
• scikit-learn>=1.3.0
• pandas>=2.0.0

Basic Usage (30 seconds to results)

from cans.config import CANSConfig
from cans.utils.data import create_sample_dataset, get_data_loaders
from cans.models import CANS
from cans.models.gnn_modules import GCN
from cans.pipeline.runner import CANSRunner
from transformers import BertModel
import torch

# 1. Create configuration
config = CANSConfig()
config.training.epochs = 10

# 2. Load data (or create sample data)
datasets = create_sample_dataset(n_samples=1000, n_features=64)
train_loader, val_loader, test_loader = get_data_loaders(datasets, batch_size=32)

# 3. Create model
gnn = GCN(in_dim=64, hidden_dim=128, output_dim=256)
bert = BertModel.from_pretrained("bert-base-uncased")
model = CANS(gnn, bert, fusion_dim=256)

# 4. Train
optimizer = torch.optim.AdamW(model.parameters(), lr=0.001)
runner = CANSRunner(model, optimizer, config)
history = runner.fit(train_loader, val_loader)

# 5. Evaluate
results = runner.evaluate(test_loader)
print(f"Test MSE: {results['mse']:.4f}")
print(f"Average Treatment Effect: {results['ate']:.4f}")

🖥️ Command Line Interface (CLI)

CANS provides six powerful CLI commands for complete causal analysis workflow:

💡 New to CANS? Check out our comprehensive User Guide for step-by-step tutorials, troubleshooting, and best practices!

1. Validate Causal Assumptions (cans-validate)

Test critical causal assumptions before modeling:

# Basic usage
cans-validate --data data.csv --treatment intervention --outcome conversion_rate

# Specify features explicitly  
cans-validate --data marketing_data.csv \
              --treatment campaign_type \
              --outcome revenue \
              --features age,income,education,region \
              --output validation_results.json \
              --verbose

# Example output:
# {
#   "unconfoundedness_test": {
#     "valid": true,
#     "p_value": 0.23,
#     "method": "backdoor_criterion"
#   },
#   "positivity_test": {
#     "valid": true,
#     "overlap_score": 0.85,
#     "min_propensity": 0.05
#   },
#   "sutva_test": {
#     "valid": true,
#     "interference_score": 0.02
#   }
# }

2. Evaluate Model Performance (cans-evaluate)

Assess causal model predictions:

# Evaluate predictions file with columns: mu0, mu1, treatments, outcomes
cans-evaluate --predictions model_predictions.csv --format json

# Save detailed evaluation report
cans-evaluate --predictions predictions.csv \
              --output evaluation_report.txt \
              --format text

# Example output metrics:
# - Average Treatment Effect (ATE): 2.34 ± 0.18
# - Factual MSE: 0.045
# - PEHE (Precision in Estimation of Heterogeneous Effects): 0.12
# - Individual Treatment Effect R²: 0.73

3. Complete Causal Analysis (cans-analyze)

Run end-to-end causal inference workflow:

# Quick analysis with default configuration
cans-analyze --data patient_data.csv --output-dir results/

# Use custom configuration
cans-analyze --data social_media.csv \
             --config custom_config.json \
             --output-dir social_analysis/

# Creates structured output:
# results/
# ├── assumptions_validation.json
# ├── model_performance.json  
# ├── causal_effects_summary.json
# ├── individual_effects.csv
# └── analysis_report.html

4. Interactive Tutorials (cans-tutorials)

Set up hands-on tutorials in your workspace:

# Create tutorial workspace in current directory
cans-tutorials

# Specify custom directory
cans-tutorials --dir my-learning-space

# List available tutorials
cans-tutorials --list

# Creates structured tutorial environment:
# cans-tutorials/
# ├── README.md                        # Quick start guide
# ├── TUTORIALS.md                     # Complete documentation
# ├── tutorial_01_first_steps.py       # Basic introduction
# ├── tutorial_02_data_understanding.py # Advanced data handling
# └── examples/
#     └── sample_marketing_data.csv    # Practice dataset

Tutorial Features:

• 🎯 Interactive Learning: Step-by-step Python scripts with detailed explanations
• 📊 Sample Data: Real-world-like datasets for immediate practice
• 🔬 Progressive Complexity: From basics to advanced causal inference
• 💡 Best Practices: Learn proper workflow and methodology
• 🛠️ Editable Copies: Modify and experiment with tutorial code

5. API Server (cans-server)

Start production-ready REST API:

# Start API server
cans-server
# Interactive docs at: http://localhost:8000/docs

6. LLM Integration (cans-mcp)

Enable AI assistants to perform causal analysis:

# Start MCP server for LLM integration
cans-mcp
# Enables Claude, GPT, and other LLMs to use CANS autonomously

CLI Configuration Files

Create reusable configuration files for complex analyses:

{
  "model": {
    "gnn_type": "GCN",
    "gnn_hidden_dim": 128,
    "fusion_dim": 256,
    "text_model": "distilbert-base-uncased"
  },
  "training": {
    "learning_rate": 0.001,
    "batch_size": 64,
    "epochs": 50,
    "loss_type": "cfr"
  },
  "data": {
    "graph_construction": "knn",
    "knn_k": 5,
    "scale_node_features": true
  }
}

📊 Usage Examples

Example 1: CSV Data with Real Causal Inference

from cans.utils.data import load_csv_dataset
from cans.config import CANSConfig, DataConfig

# Configure data processing
config = CANSConfig()
config.data.graph_construction = "knn"  # or "similarity" 
config.data.knn_k = 5
config.data.scale_node_features = True

# Load your CSV data
datasets = load_csv_dataset(
    csv_path="your_data.csv",
    text_column="review_text",        # Column with text data
    treatment_column="intervention",   # Binary treatment (0/1)
    outcome_column="conversion_rate",  # Continuous outcome  
    feature_columns=["age", "income", "education"],  # Numerical features
    config=config.data
)

train_dataset, val_dataset, test_dataset = datasets

# Check data quality
stats = train_dataset.get_statistics()
print(f"Treatment proportion: {stats['treatment_proportion']:.3f}")
print(f"Propensity overlap valid: {stats['propensity_overlap_valid']}")

Example 2: Advanced Configuration & Experiment Tracking

from cans.config import CANSConfig

# Create detailed configuration
config = CANSConfig()

# Model configuration
config.model.gnn_type = "GCN"
config.model.gnn_hidden_dim = 256
config.model.fusion_dim = 512
config.model.text_model = "distilbert-base-uncased"  # Faster BERT variant

# Training configuration  
config.training.learning_rate = 0.001
config.training.batch_size = 64
config.training.epochs = 50
config.training.early_stopping_patience = 10
config.training.gradient_clip_norm = 1.0
config.training.loss_type = "huber"  # Robust to outliers

# Experiment tracking
config.experiment.experiment_name = "healthcare_causal_analysis"
config.experiment.save_every_n_epochs = 5
config.experiment.log_level = "INFO"

# Save configuration for reproducibility
config.save("experiment_config.json")

# Later: load and use
loaded_config = CANSConfig.load("experiment_config.json")

Example 3: Counterfactual Analysis & Treatment Effects

from cans.utils.causal import simulate_counterfactual
import numpy as np

# After training your model...
runner = CANSRunner(model, optimizer, config)
runner.fit(train_loader, val_loader)

# Comprehensive evaluation
test_metrics = runner.evaluate(test_loader)
print("Performance Metrics:")
for metric, value in test_metrics.items():
    print(f"  {metric}: {value:.4f}")

# Counterfactual analysis
cf_control = simulate_counterfactual(model, test_loader, intervention=0)
cf_treatment = simulate_counterfactual(model, test_loader, intervention=1)

# Calculate causal effects
ate = np.mean(cf_treatment) - np.mean(cf_control)
print(f"\nCausal Analysis:")
print(f"Average Treatment Effect (ATE): {ate:.4f}")
print(f"Expected outcome under control: {np.mean(cf_control):.4f}")
print(f"Expected outcome under treatment: {np.mean(cf_treatment):.4f}")

# Individual treatment effects
individual_effects = np.array(cf_treatment) - np.array(cf_control)
print(f"Treatment effect std: {np.std(individual_effects):.4f}")
print(f"% benefiting from treatment: {(individual_effects > 0).mean()*100:.1f}%")

Example 4: Custom Data Pipeline

from cans.utils.preprocessing import DataPreprocessor, GraphBuilder
from cans.config import DataConfig
import pandas as pd

# Custom preprocessing pipeline
config = DataConfig()
config.graph_construction = "similarity"
config.similarity_threshold = 0.7
config.scale_node_features = True

preprocessor = DataPreprocessor(config)

# Process your DataFrame  
df = pd.read_csv("social_media_posts.csv")
dataset = preprocessor.process_tabular_data(
    data=df,
    text_column="post_content",
    treatment_column="fact_check_label",
    outcome_column="share_count",
    feature_columns=["user_followers", "post_length", "sentiment_score"],
    text_model="bert-base-uncased",
    max_text_length=256
)

# Split with custom ratios
train_ds, val_ds, test_ds = preprocessor.split_dataset(
    dataset, 
    train_size=0.7, 
    val_size=0.2, 
    test_size=0.1
)

🧪 Testing & Development

# Run all tests
pytest tests/ -v

# Run specific test categories  
pytest tests/test_models.py -v        # Model tests
pytest tests/test_validation.py -v    # Validation tests
pytest tests/test_pipeline.py -v      # Training pipeline tests

# Run with coverage
pytest tests/ --cov=cans --cov-report=html

# Run example scripts
python examples/enhanced_usage_example.py
python examples/enhanced_causal_analysis_example.py

📁 Framework Structure

cans-framework/
├── cans/
│   ├── __init__.py              # Main imports
│   ├── config.py                # ✨ Configuration management
│   ├── exceptions.py            # ✨ Custom exceptions
│   ├── validation.py            # ✨ Data validation utilities
│   ├── models/
│   │   ├── cans.py             # Core CANS model (enhanced)
│   │   └── gnn_modules.py      # GNN implementations
│   ├── pipeline/
│   │   └── runner.py           # ✨ Enhanced training pipeline
│   └── utils/
│       ├── causal.py           # Counterfactual simulation
│       ├── data.py             # ✨ Enhanced data loading
│       ├── preprocessing.py     # ✨ Advanced preprocessing
│       ├── logging.py          # ✨ Structured logging
│       └── checkpointing.py    # ✨ Model checkpointing
├── tests/                       # ✨ Comprehensive test suite
├── examples/                    # Usage examples
└── CLAUDE.md                   # Development guide

✨ = New/Enhanced in v2.0

🎯 Use Cases & Applications

Healthcare & Medical

# Analyze treatment effectiveness with patient records + clinical notes
datasets = load_csv_dataset(
    csv_path="patient_outcomes.csv",
    text_column="clinical_notes",
    treatment_column="medication_type", 
    outcome_column="recovery_score",
    feature_columns=["age", "bmi", "comorbidities"]
)

Marketing & A/B Testing

# Marketing campaign effectiveness with customer profiles + ad content
datasets = load_csv_dataset(
    csv_path="campaign_data.csv", 
    text_column="ad_content",
    treatment_column="campaign_variant",
    outcome_column="conversion_rate",
    feature_columns=["customer_ltv", "demographics", "behavior_score"]
)

Social Media & Content Moderation

# Impact of content moderation on engagement
datasets = load_csv_dataset(
    csv_path="posts_data.csv",
    text_column="post_text", 
    treatment_column="moderation_action",
    outcome_column="engagement_score",
    feature_columns=["user_followers", "post_length", "sentiment"]
)

🔬 Research & Methodology

CANS implements state-of-the-art causal inference techniques:

• Counterfactual Regression Networks (CFRNet): Learn representations that minimize treatment assignment bias
• Gated Fusion: Adaptively combine graph-structured and textual information
• Balanced Representation: Minimize distributional differences between treatment groups
• Propensity Score Validation: Automatic overlap checking for reliable causal estimates

Key Papers:

• Shalit et al. "Estimating individual treatment effect: generalization bounds and algorithms" (ICML 2017)
• Yao et al. "Representation learning for treatment effect estimation from observational data" (NeurIPS 2018)

🚀 Performance & Scalability

• Memory Efficient: Optimized batch processing and gradient checkpointing
• GPU Acceleration: Full CUDA support with automatic device selection
• Parallel Processing: Multi-core data loading and preprocessing
• Production Ready: Comprehensive error handling and logging

Benchmarks (approximate, hardware-dependent):

• Small: 1K samples, 32 features → ~30 sec training
• Medium: 100K samples, 128 features → ~10 min training
• Large: 1M+ samples → Scales with batch size and hardware

🌐 API & Integration

CANS provides comprehensive API access for integration with web applications, services, and AI systems:

REST API Server

# Start REST API server
cans-server
# or
uvicorn cans.api.server:app --host 0.0.0.0 --port 8000

# Interactive docs at: http://localhost:8000/docs

MCP Server for LLMs

# Start MCP server for LLM integration
cans-mcp
# Enables LLMs to directly perform causal analysis

Python API Client

from cans.api.client import CANSAPIClient

client = CANSAPIClient(api_key="your-key")
results = client.validate_assumptions(
    data="data.csv",
    treatment_column="treatment",
    outcome_column="outcome"
)

🔗 API Features:

• RESTful endpoints for all CANS functionality
• Model Context Protocol (MCP) server for LLM integration
• Authentication and rate limiting
• Async/await support
• Comprehensive error handling
• Interactive documentation

📖 Learn More: API Guide

📚 Documentation & Resources

📚 Documentation Hierarchy

1. Interactive Tutorials (cans-tutorials) - Hands-on learning with guided examples and sample data
2. Getting Started - 5-minute quickstart tutorial
3. User Guide - Comprehensive guide with tutorials, best practices, and FAQ
4. API Guide - Complete API integration guide with examples
5. README - This file with complete feature overview
6. Examples - Real-world use cases and workflows
7. Changelog - Version history and updates

🔧 For Developers

• Tests - 100+ unit tests and usage patterns
• Configuration - Project setup and dependencies
• API Reference - In-code documentation with detailed docstrings

🆘 Getting Help

• New to CANS? → Run cans-tutorials for hands-on interactive learning
• First time? → Getting Started Guide
• Need detailed help? → User Guide with FAQ and troubleshooting
• Found a bug? → GitHub Issues
• Have questions? → Email durai@infinidatum.net

🤝 Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Add tests for new functionality
4. Run tests: pytest tests/ -v
5. Submit a pull request

Areas we'd love help with:

• Additional GNN architectures (GraphSAGE, Graph Transformers)
• More evaluation metrics for causal inference
• Integration with popular ML platforms (MLflow, Weights & Biases)
• Performance optimizations

👨‍🔬 Authors

Durai Rajamanickam – @duraimuruganr reach out to durai@infinidatum.net

📜 License

CANS Framework Custom License - Free for academic and personal use, commercial use requires permission.

• ✅ Academic & Research Use: Freely use for educational and non-commercial research
• ✅ Personal Projects: Use for personal and non-commercial experimentation
• 📋 Commercial Use: Requires written permission from durai@infinidatum.net

See LICENSE for full terms and conditions.

---

Ready to get started? Try the 30-second quick start above, or dive into the detailed examples! 🚀