nexus-stats 6.0.1

Fixed-memory, zero-allocation streaming statistics for real-time systems
Documentation
# Type Matrix

Full type availability for all algorithms. ✓ = available, — = not applicable.

## Numeric Types

| Algorithm | f32 | f64 | i32 | i64 | i128 |
|-----------|:---:|:---:|:---:|:---:|:----:|
| **Change Detection** | | | | | |
| CUSUM ||||||
| MOSUM ||||||
| ShiryaevRoberts ||||||
| MultiGate ||||||
| RobustZScore ||||||
| AdaptiveThreshold ||||||
| **Smoothing** | | | | | |
| EMA ||||||
| AsymmetricEMA ||||||
| KAMA ||||||
| Kalman1D ||||||
| Holt ||||||
| Spring ||||||
| SlewLimiter ||||||
| WindowedMedian ||||||
| **Statistics** | | | | | |
| Welford ||||||
| Moments ||||||
| EwmaVariance ||||||
| Covariance ||||||
| HarmonicMean ||||||
| **Regression** | | | | | |
| LinearRegression ||||||
| EwLinearRegression ||||||
| PolynomialRegression ||||||
| EwPolynomialRegression ||||||
| ExponentialRegression ||||||
| LogarithmicRegression ||||||
| PowerRegression ||||||
| **Bayesian Inference** | | | | | |
| BetaBinomial ||||||
| GammaPoisson ||||||
| **Hypothesis Testing** | | | | | |
| SprtBernoulli ||||||
| SprtGaussian ||||||
| **Adaptive Filters** | | | | | |
| LmsFilter ||||||
| NlmsFilter ||||||
| RlsFilter ||||||
| LogisticRegression ||||||
| OnlineKMeans ||||||
| **State Estimation** | | | | | |
| Kalman2D ||||||
| Kalman3D ||||||
| **Optimization** | | | | | |
| OnlineGD ||||||
| AdaGrad ||||||
| Adam ||||||
| **Signal Analysis** | | | | | |
| Autocorrelation *(alloc)* ||||||
| CrossCorrelation *(alloc)* ||||||
| **Information Theory** | | | | | |
| Entropy *(alloc)* ||||||
| TransferEntropy ||||||
| **Monitoring** | | | | | |
| Drawdown ||||||
| RunningMin/Max ||||||
| WindowedMin/Max ||||||
| PeakHoldDecay ||||||
| MaxGauge ||||||
| Liveness ||||||
| EventRate ||||||
| CoDel ||||||
| Saturation ||||||
| ErrorRate ||||||
| TrendAlert ||||||
| Jitter ||||||
| **Frequency** | | | | | |
| TopK | generic key | | u64 count | | |
| FlexProportion ||||||
| DecayAccum ||||||
| **Utilities** | | | | | |
| Debounce ||| u32 | u64 ||
| DeadBand ||||||
| Hysteresis ||||||
| BoolWindow ||| runtime-sized | | |
| PeakDetector ||||||
| LevelCrossing ||||||
| FirstDiff ||||||
| SecondDiff ||||||

## Integer vs Float Decision

**Use integer types when:**
- Signal is naturally integer (nanoseconds, ticks, counts)
- No floating point is available (embedded `no_std`)
- You want deterministic results (no float rounding)

**Use float types when:**
- Signal is naturally continuous (prices, rates, percentages)
- Algorithm requires division or transcendentals (Welford, Kalman, Holt)
- You need exponential smoothing with fine-grained alpha

## Feature-Gated Types

| Feature | What it enables |
|---------|----------------|
| `std` (default) | Hardware `sqrt`/`exp`/`ln` for Welford, ShiryaevRoberts, CrossCorrelation, Entropy, etc. |
| `libm` | Pure Rust `sqrt`/`exp`/`ln` for `no_std` environments |
| `alloc` | Runtime-sized windows (MOSUM, WindowedMedian, KAMA, BoolWindow), heap tables (TransferEntropy), and runtime-sized buffers (Autocorrelation, CrossCorrelation, Entropy) |