foxhunt/ml/examples/validate_225_features_runtime.rs

//! Runtime Validation Script for 225-Feature Extraction
//!
//! Wave 3 Agent 28: Feature Dimension Runtime Validation
//!
//! This script validates:
//! 1. Feature vector shape is (N, 225) where N ≤ 100
//! 2. Warmup period handling (50 bars)
//! 3. No NaN or Inf values in output
//! 4. All feature indices are populated correctly
//!
//! Usage:
//!   cargo run --example validate_225_features_runtime --release

use anyhow::{Context, Result};
use chrono::{Duration, Utc};
use ml::features::extraction::{extract_ml_features, OHLCVBar};
use std::time::Instant;

fn main() -> Result<()> {
    println!("🔍 Wave 3 Agent 28: 225-Feature Runtime Validation");
    println!("{}", "=".repeat(70));
    println!();

    // Step 1: Create synthetic OHLCV data (100 bars)
    println!("📊 Step 1: Creating 100 synthetic OHLCV bars...");
    let bars = create_synthetic_bars(100)?;
    println!("✓ Created {} OHLCV bars", bars.len());
    println!();

    // Step 2: Extract features and measure performance
    println!("🔬 Step 2: Extracting 225-dimensional features...");
    let start = Instant::now();
    let features =
        extract_ml_features(&bars).context("Failed to extract 225-dimensional features")?;
    let duration = start.elapsed();
    println!(
        "✓ Extracted {} feature vectors in {:.3}ms",
        features.len(),
        duration.as_secs_f64() * 1000.0
    );
    println!(
        "  Average: {:.3}μs per bar",
        duration.as_micros() as f64 / features.len() as f64
    );
    println!();

    // Step 3: Verify output shape
    println!("📐 Step 3: Verifying feature dimensions...");
    let expected_vectors = bars.len() - 50; // 100 bars - 50 warmup = 50 vectors
    println!("  Input bars: {}", bars.len());
    println!("  Warmup period: 50 bars");
    println!("  Expected vectors: {} (100 - 50)", expected_vectors);
    println!("  Actual vectors: {}", features.len());

    if features.len() == expected_vectors {
        println!("✓ Feature vector count is CORRECT (N = {})", features.len());
    } else {
        println!("❌ Feature vector count MISMATCH!");
        println!("   Expected: {}, Got: {}", expected_vectors, features.len());
        anyhow::bail!("Feature vector count validation failed");
    }

    if !features.is_empty() && features[0].len() == 225 {
        println!("✓ Feature dimension is CORRECT (225 per vector)");
    } else {
        println!("❌ Feature dimension MISMATCH!");
        if !features.is_empty() {
            println!("   Expected: 225, Got: {}", features[0].len());
        }
        anyhow::bail!("Feature dimension validation failed");
    }
    println!();

    // Step 4: Check for NaN/Inf values
    println!("🔍 Step 4: Validating feature values (NaN/Inf check)...");
    let mut nan_count = 0;
    let mut inf_count = 0;
    let mut total_features = 0;

    for (vec_idx, feature_vec) in features.iter().enumerate() {
        for (feat_idx, &value) in feature_vec.iter().enumerate() {
            total_features += 1;
            if value.is_nan() {
                nan_count += 1;
                if nan_count <= 5 {
                    println!("  ⚠ NaN at vector[{}], feature[{}]", vec_idx, feat_idx);
                }
            }
            if value.is_infinite() {
                inf_count += 1;
                if inf_count <= 5 {
                    println!("  ⚠ Inf at vector[{}], feature[{}]", vec_idx, feat_idx);
                }
            }
        }
    }

    if nan_count == 0 && inf_count == 0 {
        println!("✓ All {} features are VALID (no NaN/Inf)", total_features);
    } else {
        println!("❌ INVALID features detected:");
        println!(
            "   NaN count: {} ({:.2}%)",
            nan_count,
            (nan_count as f64 / total_features as f64) * 100.0
        );
        println!(
            "   Inf count: {} ({:.2}%)",
            inf_count,
            (inf_count as f64 / total_features as f64) * 100.0
        );
        anyhow::bail!("Feature validation failed: NaN or Inf values detected");
    }
    println!();

    // Step 5: Verify warmup period behavior
    println!("🕐 Step 5: Verifying warmup period behavior...");

    // Test with exactly 50 bars (should fail)
    let warmup_bars = create_synthetic_bars(50)?;
    let warmup_result = extract_ml_features(&warmup_bars);
    match warmup_result {
        Ok(_) => {
            println!("❌ Should have failed with 50 bars (warmup period)");
            anyhow::bail!("Warmup validation failed: extracted features from 50 bars");
        },
        Err(e) => {
            println!("✓ Correctly rejects 50 bars: {}", e);
        },
    }

    // Test with 51 bars (should succeed with 1 vector)
    let minimal_bars = create_synthetic_bars(51)?;
    let minimal_result = extract_ml_features(&minimal_bars)?;
    if minimal_result.len() == 1 {
        println!("✓ Correctly extracts 1 vector from 51 bars (51 - 50 warmup)");
    } else {
        println!(
            "❌ Expected 1 vector from 51 bars, got {}",
            minimal_result.len()
        );
        anyhow::bail!("Warmup validation failed: incorrect vector count");
    }
    println!();

    // Step 6: Feature range analysis
    println!("📊 Step 6: Feature range analysis (first 10 features)...");
    if !features.is_empty() {
        let first_vec = &features[0];
        for i in 0..10.min(first_vec.len()) {
            let value = first_vec[i];
            println!("  Feature[{}]: {:.6}", i, value);
        }
    }
    println!();

    // Final summary
    println!("{}", "=".repeat(70));
    println!("🎉 VALIDATION SUMMARY");
    println!("{}", "=".repeat(70));
    println!(
        "✓ Feature vector count: {} (N = 100 bars - 50 warmup)",
        features.len()
    );
    println!("✓ Feature dimensions: 225 per vector");
    println!("✓ NaN/Inf check: PASSED (0 invalid values)");
    println!("✓ Warmup period: CORRECT (50 bars)");
    println!(
        "✓ Performance: {:.3}μs per bar (target: <1000μs)",
        duration.as_micros() as f64 / features.len() as f64
    );
    println!();
    println!("🚀 225-Feature extraction system is PRODUCTION READY!");
    println!();

    Ok(())
}

/// Create synthetic OHLCV bars for testing
fn create_synthetic_bars(count: usize) -> Result<Vec<OHLCVBar>> {
    let mut bars = Vec::with_capacity(count);
    let base_time = Utc::now();
    let base_price = 4500.0; // ES.FUT-like price

    for i in 0..count {
        let timestamp = base_time + Duration::minutes(i as i64);

        // Create realistic price movement (random walk with drift)
        let price_delta = (i as f64 * 0.5).sin() * 2.0; // Oscillating trend
        let close = base_price + price_delta;
        let high = close + (i as f64 * 0.1).sin().abs() * 1.5;
        let low = close - (i as f64 * 0.1).cos().abs() * 1.5;
        let open = close - price_delta * 0.3;

        // Realistic volume (1000-5000 contracts)
        let volume = 2000.0 + (i as f64 * 0.2).cos() * 1500.0;

        bars.push(OHLCVBar {
            timestamp,
            open,
            high,
            low,
            close,
            volume: volume.abs(),
        });
    }

    Ok(bars)
}