unsupervised.py

import pandas as pd
import numpy as np
from sklearn.preprocessing import StandardScaler, LabelEncoder
from sklearn.cluster import KMeans
from sklearn.metrics import silhouette_score
import matplotlib.pyplot as plt
import seaborn as sns

# Set the style using seaborn
sns.set_style("whitegrid")  # or "darkgrid", "white", "dark", "ticks"

def analyze_healthcare_patterns():
    """Analyze healthcare patterns to support SDG 3 goals."""
    try:
        # Load and preprocess data with ethical considerations
        df = pd.read_csv('archive/healthcare_dataset.csv')
        
        # Reduce dataset size by sampling
        df = df.sample(n=min(500, len(df)), random_state=42)
        
        print("SDG 3 Healthcare Analysis Initiative")
        print("===================================")
        print(f"Analyzing {len(df)} patient records for healthcare patterns\n")
        
        # Optimize memory usage by selecting only necessary columns
        df = df[['Name', 'Age', 'Billing Amount', 'Medical Condition', 
                 'Admission Type']].copy()
        
        # Convert data types to reduce memory
        df['Age'] = df['Age'].astype('int32')
        df['Billing Amount'] = df['Billing Amount'].astype('float32')
        
        # Ethical consideration: Anonymize patient data
        df['Name'] = 'ANONYMIZED'
        
        # Select features relevant to healthcare access analysis
        features = ['Age', 'Billing Amount', 'Medical Condition', 'Admission Type']
        X = df[features].copy()
        
        # Preprocess categorical variables
        le = LabelEncoder()
        X['Medical Condition'] = le.fit_transform(X['Medical Condition'])
        X['Admission Type'] = le.fit_transform(X['Admission Type'])
        
        # Scale numerical features
        scaler = StandardScaler()
        X_scaled = scaler.fit_transform(X)
        
        # Find optimal clusters using silhouette analysis
        optimal_clusters = find_optimal_clusters(X_scaled)
        
        # Apply clustering
        kmeans = KMeans(n_clusters=optimal_clusters, random_state=42)
        df['Cluster'] = kmeans.fit_predict(X_scaled)
        
        # Generate visualizations
        create_healthcare_visualizations(df)
        
        # Generate insights
        generate_healthcare_insights(df)
        
    except Exception as e:
        print(f"Analysis Error: {str(e)}")
    finally:
        plt.close('all')

def find_optimal_clusters(X_scaled):
    """Determine optimal number of patient clusters."""
    silhouette_scores = []
    for k in range(2, 7):
        kmeans = KMeans(n_clusters=k, random_state=42)
        labels = kmeans.fit_predict(X_scaled)
        score = silhouette_score(X_scaled, labels)
        silhouette_scores.append(score)
    return range(2, 7)[np.argmax(silhouette_scores)]

def create_healthcare_visualizations(df):
    """Create visualizations for healthcare analysis."""
    # Use a more efficient backend
    plt.switch_backend('Agg')
    
    # Reduce figure quality for faster rendering
    plt.style.use('default')  # More efficient than seaborn style
    fig, axes = plt.subplots(2, 1, figsize=(10, 8), dpi=80)  # Reduced size and DPI
    
    # Healthcare Access Pattern Visualization
    scatter = axes[0].scatter(df['Age'], df['Billing Amount'], 
                            c=df['Cluster'], cmap='viridis', 
                            alpha=0.6, s=50)
    axes[0].set_xlabel('Patient Age')
    axes[0].set_ylabel('Healthcare Cost ($)')
    axes[0].set_title('Healthcare Access and Cost Patterns')
    plt.colorbar(scatter, ax=axes[0], label='Patient Group')
    
    # Medical Condition Distribution
    condition_dist = pd.crosstab(df['Cluster'], df['Medical Condition'])
    condition_dist.plot(kind='bar', stacked=True, ax=axes[1])
    axes[1].set_title('Healthcare Needs by Patient Group')
    axes[1].set_xlabel('Patient Group')
    axes[1].set_ylabel('Number of Patients')
    
    plt.tight_layout()
    plt.show()

def generate_healthcare_insights(df):
    """Generate insights for healthcare policy recommendations."""
    print("\nHealthcare Access Insights:")
    print("==========================")
    for cluster in df['Cluster'].unique():
        cluster_data = df[df['Cluster'] == cluster]
        print(f"\nPatient Group {cluster + 1}:")
        print(f"- Average Age: {cluster_data['Age'].mean():.1f} years")
        print(f"- Average Healthcare Cost: ${cluster_data['Billing Amount'].mean():.2f}")
        print(f"- Most Common Condition: {cluster_data['Medical Condition'].mode()[0]}")

if __name__ == "__main__":
    analyze_healthcare_patterns()