Lab: Feature Pipeline

Build a SIMD-accelerated feature computation pipeline.

Objectives

• Compute feature statistics
• Implement normalization transforms
• Build a composable pipeline

Demo Code

See demos/course3/week2/feature-pipeline/

Lab Exercise

See labs/course3/week2/lab_2_5_feature_pipeline.py

Key Transforms

#![allow(unused)]
fn main() {
pub fn normalize_zscore(values: &[f32]) -> Result<Vec<f32>, FeatureError> {
    let stats = compute_statistics(values)?;
    Ok(values.iter()
        .map(|v| (v - stats.mean) / stats.std_dev)
        .collect())
}

pub fn normalize_minmax(values: &[f32]) -> Result<Vec<f32>, FeatureError> {
    let stats = compute_statistics(values)?;
    let range = stats.max - stats.min;
    Ok(values.iter()
        .map(|v| (v - stats.min) / range)
        .collect())
}
}

Validation

Run tests:

cd demos/course3/week2/feature-pipeline
cargo test