statsrust is a high-performance statistical analysis library for Rust developers, designed with emphasis on numerical stability, mathematical correctness, and user-friendly API.
- Descriptive Statistics: Comprehensive implementation of measures of central tendency, position, and variability
- Kernel Density Estimation: Multiple kernel functions with efficient implementation
- Normal Distribution Model: Complete algebraic operations on normal distributions
- Numerical Stability: Carefully designed algorithms to prevent overflow, underflow, and cancellation errors
- Comprehensive Error Handling: Detailed error messages for edge cases
- Generic Input Support: Works with various numeric types
Add this to your Cargo.toml:
[dependencies]
statsrust = "0.1.0" # Replace with the latest versionuse statsrust::*;
fn main() -> Result<(), StatError> {
let data = vec![1.0, 2.0, 3.0, 4.0, 5.0];
// Basic descriptive statistics
let mean = mean(&data)?;
let median = median(&data)?;
let variance = variance(&data, None)?;
println!("Mean: {:.2}, Median: {:.2}, Variance: {:.2}", mean, median, variance);
// Kernel Density Estimation
let kde = kde(&data, 0.5, "normal", false)?;
println!("Density at 3.0: {:.4}", kde(3.0));
// Normal distribution operations
let dist = NormalDist::from_samples(&data)?;
println!("Distribution: N(μ={:.2}, σ={:.2})", dist.mean(), dist.stdev());
Ok(())
}Our algorithms are designed to avoid common numerical issues:
- Geometric mean uses logarithmic transformation to prevent overflow
- Variance calculation uses centered data to avoid catastrophic cancellation
- Inverse CDF approximations maintain precision while balancing performance
Efficient implementation with multiple kernel functions:
- Gaussian, Epanechnikov, Triangular, Quartic, Triweight, and more
- Optimized for bounded kernels using binary search
- Supports both PDF and CDF estimation
- Random sample generation from estimated distributions
Complete algebraic operations:
let dist1 = NormalDist::new(0.0, 1.0)?;
let dist2 = NormalDist::new(1.0, 2.0)?;
// Distribution operations
let sum_dist = dist1 + dist2; // N(1.0, √5)
let scaled_dist = dist1 * 2.0; // N(0.0, 2.0)
let overlap = dist1.overlap(&dist2); // Calculate distribution overlapFor comprehensive documentation:
We welcome contributions! Please see our Contribution Guidelines for details on how to report bugs, suggest enhancements, or submit pull requests.
This project is dual-licensed under:
Documentation content is licensed under:
This document is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
Original author: statsrust Authors