Crate asap_rs

ASAP-RS

asap-rs is a Rust implementation of Automatic Smoothing for Attention Prioritization (ASAP), an algorithm for automatically smoothing time series data to improve visualization.

The algorithm is based on the paper “ASAP: Prioritizing Attention via Time Series Smoothing” by Kexin Rong, Peter Bailis, et al. from Stanford University.

Features

• Automatically determines optimal smoothing window size based on data characteristics
• Maintains important patterns and anomalies while reducing noise
• Preserves kurtosis to ensure the overall distribution shape is maintained
• Optimizes for reduced visual roughness
• Special handling for different data patterns (trending, periodic, zigzag)

Usage

use asap_rs::smooth;
 
fn main() {
    // Example time series data
    let data = vec![1.0, 2.0, 1.5, 3.0, 2.5, 4.0, 3.5, 5.0, 4.5, 6.0];
     
    // Apply ASAP smoothing with target resolution of 5 points
    let smoothed_data = smooth(&data, 5);
     
    println!("Original data: {:?}", data);
    println!("Smoothed data: {:?}", smoothed_data);
}

Algorithm Overview

ASAP automatically finds the optimal window size for smoothing by:

1. Computing the roughness (standard deviation of differences between consecutive points)
2. Ensuring that kurtosis (measure of “tailedness” of distribution) is preserved
3. Using autocorrelation to identify potential window sizes aligned with patterns in the data
4. Selecting the window size that minimizes roughness while preserving key statistical properties

Pattern-Specific Processing

ASAP-RS includes specialized handling for different types of time series patterns:

• Periodic data: Windows aligned with the period are prioritized to preserve seasonality
• Trend data: Appropriate windows are selected to maintain the trend while removing noise
• Zigzag patterns: Special detection and handling for highly alternating data patterns

For zigzag patterns (rapidly alternating high-low values), ASAP uses a relaxed kurtosis constraint and prioritizes a window size of 2, which is mathematically optimal for smoothing such patterns while preserving their mean value.

Resolution Parameter

The resolution parameter controls the trade-off between smoothing and detail preservation:

• Lower values (e.g., 2-5) apply more aggressive smoothing, highlighting major trends
• Higher values (e.g., 50+) preserve more detail, suitable for dense visualizations

The parameter roughly corresponds to the desired number of points in the output, though ASAP may return more or fewer points based on what best preserves important patterns.

Re-exports

pub use smoothing::adaptive_smoothing::smooth;

Modules

fft

smoothing

statistics

utils