Expand description
A lightweight library that provides some basic parametric and non-parametric statistics and hypothesis tests.
This library strives for ease of use and small size. The only numeric types used are u64 and f64. Functions in this library operate on primitive types, iterators, or slices.
There is no support for parallel processing. In particular, there is no dependence on large libraries like polars that support parallel processing on custom data structures.
§Cargo features
By default, use of this library as a dependency includes modules core, normal, binomial, and wilcoxon. The aok module is not included by default.
Each module other than core (which is always enabled) has an associated cargo feature that enables the module. To include only selected modules, specify default-features = false in the dependency declaration (or --no-default-features on the command line) and specify the desired features in the dependency declaration (or command line).
§Error handling
Functions in this library are designed not to panic. Although functions in crates this library depends on may panic, this library implements extensive error handling, validation, and testing to prevent panics. If you encounter a panic, it is a bug, so please create an issue to report it.
Except in a couple of specifically documented places, functions in this library only return finite (i.e., not NaN, Infinity, or -Infinity) values unless a non-finite value is provided as an input.
Modules§
- aok
- Supports the coercion of
Result<Value, E>to the underlying typeValue, producing a suitable fallback output value instead of panicking in case of error. - binomial
- Statistics related to samples of the Bernoulli distribution. The Binomial distribution with parameters
nandpis the distribution of the sum ofnindependent Bernoulli random variables with probability of successp. - core
- Core sample statistics and common types.
- normal
- Statistics related to the Normal distribution and Student’s t distribution, including t-tests.
- wilcoxon
- Statistics related to the Wilcoxon rank sum two-sample test, also known as the Mann-Whitney U test.