rstats 1.1.0

Statistical Measures, Vector Algebra, Geometric Median, Data Analysis and Machine Learning
Documentation

Rstats

Usage

Insert rstats = "^1" in the Cargo.toml file, under [dependencies].

Use in your source files any of the following structs, if needed:

use rstats::{Mstats};  
use indxvec::{MinMax};  
use medians::Med;

and any of the following rstats traits:

use rstats::{Stats, Vecu8, Vecg, MutVecg, VecVecu8, VecVec, VecVecg};

and any of the following helper functions:
use rstats::{i64tof64,wsum};

The latest (nightly) version of this readme file and everything, is always available in the github repository rstats. Sometimes it may be a little ahead of the crates.io release versions.

It is highly recommended to read and run tests/tests.rs, which shows examples of usage.

To run all the tests, use single thread in order to produce the results in the right order:
cargo test --release -- --test-threads=1 --nocapture --color always

Introduction

Rstats is primarily about characterising multidimensional sets of points, with applications to Machine Learning and Big Data Analysis. It uses non analytical statistics, where the 'random variables' are replaced by vectors of real data. Probabilities densities and other parameters are always obtained from the data, not from some assumed distributions.

This crate begins with basic statistical measures and vector algebra, which provide self-contained tools for the multidimensional algorithms but can also be used in their own right.

Our treatment of multidimensional sets of points (vectors) is constructed from the first principles. Some original concepts, not found elsewhere, are introduced and implemented here:

  • gmedian - fast multidimensional (geometric) median algorithm.

  • madgm - generalisation of robust data spread estimator known as 'MAD' in 1d (median of absolute deviations from median), to multiple dimensions (nd).

  • comediance - instead of covariance (matrix). It is obtained by supplying covar with the geometric median instead of the usual centroid. Thus zero median vectors are replacing zero mean vectors in covariance calculations.

  • median correlation- in one dimension (1d), our mediancorr method is to replace Pearson's correlation. We define median correlation as cosine of an angle between two zero median vectors (instead of Pearson's zero mean vectors).

Zero median vectors are generally preferable to the commonly used zero mean vectors.

In n dimensions (nd), many authors 'cheat' by using quasi medians (1-d medians along each axis). Quasi medians are a poor start to stable characterisation of multidimensional data. In a highly dimensional space, they are not even any faster to compute.

Specifically, all such 1d measures are sensitive to the choice of axis and thus are affected by rotation.

In contrast, analyses based on the true geometric median (gm) are axis (rotation) independent. Also, they are more stable, as medians have a 50% breakdown point (the maximum possible). They are computed here by methods gmedian and its weighted version wgmedian, in traits vecvec and vecvecg respectively.

Implementation

The main constituent parts of Rstats are its traits. The selection of traits (to import) is primarily determined by the types of objects to be handled. These are mostly vectors of arbitrary length (dimensionality). The main traits are implementing methods applicable to:

  • Stats: a single vector (of numbers),
  • Vecg: methods (of vector algebra) operating on two vectors, e.g. scalar product
  • MutVecg: some of the above methods, mutating self
  • VecVec: methods operating on n vectors,
  • VecVecg: methods for n vectors, plus another generic argument, e.g. vector of weights.

In other words, the traits and their methods operate on arguments of their required categories. In classical statistical parlance, the main categories correspond to the number of 'random variables'. However, the vectors' end types (for the actual data) are mostly generic: usually some numeric type. There are also some traits specialised for input end type u8 and some that take mutable self. End type f64 is most commonly used for the results.

Documentation

For more detailed comments, plus some examples, see the source. You may have to unclick the 'implementations on foreign types' somewhere near the bottom of the page in the rust docs to get to it. (Since these traits are implemented over the pre-existing Rust Vec type).

Struct

  • struct MStats holds the central tendency, e.g. mean, and spread, e.g. standard deviation.

Auxiliary Functions

  • i64tof64: converts an i64 vector to f64,
  • wsum: sum of a sequence 1..n, also the size of a lower/upper triangular matrix below/above the diagonal (n*(n+1)/2.).

Trait Stats

One dimensional statistical measures implemented for all numeric end types.

Its methods operate on one slice of generic data and take no arguments. For example, s.amean() returns the arithmetic mean of the data in slice s. Some of these methods are checked and will report all kinds of errors, such as an empty input. This means you have to apply to their results ?, .unwrap() or something better.

Included in this trait are:

  • means (arithmetic, geometric and harmonic),
  • standard deviations,
  • linearly weighted means (useful for time dependent data analysis),
  • probability density function (pdf)
  • autocorrelation, entropy
  • linear transformation to [0,1],
  • other measures and vector algebra operators

Note that fast implementation of 1d medians is as of version 1.1.0 in crate medians:
use medians::{Med,Median};

Trait Vecg

Generic vector algebra operations between two slices &[T], &[U] of any length (dimensionality). It may be necessary to invoke some using the 'turbofish' ::<type> syntax to indicate the type U of the supplied argument, e.g.:
datavec.as_slice().methodname::<f64>(arg)
This is because Rust is currently incapable of inferring the type ('the inference bug').

  • Vector additions, subtractions and products (scalar, kronecker, outer),
  • Other relationships and measures of difference,
  • Pearson's, Spearman's and Kendall's correlations,
  • Median correlation, which we define analogously to Pearson's, as cosine of an angle between two zero median vectors (instead of his zero mean vectors).
  • Joint pdf, joint entropy, statistical independence (based on mutual information).

This trait is unchecked (for speed), so some caution with data is advisable.

Trait MutVecg

A select few of the Stats and Vecg methods (e.g. mutable vector addition, subtraction and multiplication) are reimplemented under these traits, so that they can mutate self in-place. This is more efficient and convenient in some circumstances, such as in vector iterative methods.

Trait Vecu8

Some vector algebra as above that can be more efficient when the end type happens to be u8 (bytes). They have u8 appended to their names to avoid confusion with Vecg methods.

  • Relationships between two vectors (of bytes)
  • Frequency count of bytes by their values (histogram, pdf, jointpdf)
  • Entropy, jointentropy, independence (different algorithms to those in Vecg)

Trait VecVec

Relationships between n vectors (in d dimensions). This general data domain is denoted here as (nd). It is in nd where the main original contribution of this library lies. True geometric median (gm) is found by fast and stable iteration, using improved Weiszfeld's algorithm gmedian. This algorithm solves Weiszfeld's convergence and stability problems in the neighbourhoods of existing set points.

  • centroid, medoid, outliers, gm
  • sums of distances, radius of a point (as its distance from gm)
  • characterisation of a set of multidimensional points by the mean, standard deviation, median of its points' radii. These are useful recognition measures for the set.
  • transformation to zero geometric median data,
  • multivariate trend (regression) between two sets of nd points,
  • covariance and comediance matrices (weighted and unweighted).

Warning: trait VecVec is entirely unchecked, so check your data upfront.

Trait VecVecg

Methods which take an additional generic vector argument, such as a vector of weights for computing weighted geometric medians.

Appendix I: Terminology

Including some new definitions for sets of nd points, i.e. n points in d dimensional space

  • Centroid/Centre/Mean is the (generally non member) point that minimises the sum of squares of distances to all member points. Thus it is susceptible to outliers. Specifically, it is the n-dimensional arithmetic mean. By drawing physical analogy with gravity, it is sometimes called 'the centre of mass'. Centroid can also sometimes mean the member of the set which is the nearest to the Centre. Here we follow the common (if somewhat confusing) usage: Centroid = Centre = Arithmetic Mean.

  • Quasi/Marginal Median is the point minimising sums of distances separately in each dimension (its coordinates are 1-d medians along each axis). It is a mistaken concept which we do not use here.

  • Tukey Median is the point maximising Tukey's Depth, which is the minimum number of (outlying) points found in a hemisphere in any direction. Potentially useful concept but only partially implemented here by tukeyvec, as its advantages over the geometric median are not clear.

  • Median or the true geometric median (gm), is the point (generally non member), which minimises the sum of distances to all members. This is the one we want. It is much less susceptible to outliers than centroid. In addition, unlike quasi median, gm is rotation independent.

  • Medoid is the member of the set with the least sum of distances to all other members. Equivalently, the member which is the nearest to the gm.

  • Outlier is the member of the set with the greatest sum of distances to all other members. Equivalently, it is the point furthest from the gm.

  • Zero median vectors are obtained by subtracting the gm (placing the origin of the coordinate system at the gm). This is a proposed alternative to the commonly used zero mean vectors, obtained by subtracting the centroid.

  • MADGM (median of distances from gm). This is a generalisation of MAD (median of absolute differences) measure from 1d to nd. It is a robust measure of data spread.

  • Comediance is similar to covariance, except that zero median vectors are used to compute it, instead of zero mean vectors.

  • Median correlation between two vectors. We define it analogously to Pearson, as cosine of an angle between two 'normalised' vectors. Pearson 'normalises' by subtracting the mean from all components, we subtract the median.

Appendix II: Recent Releases

  • Version 1.1.0 - Big release. Added dependency on crate medians for fast 1D medians. Simplifications: subsumed module mutstats.rs into mutvec.rs. Removed traits Mutstats and MutVecf64 and added their few methods to trait MutVecg. Added some more doc comments here. Generalisations: methods in Vecg and MutVecg now work on any type T of self and a potentially different type U for their argument. They should be called with the 'turbofish' syntax.

  • Version 1.0.21 - Changed imports from indxvec to fit with its latest multicoloured version 1.2.1.

  • Version 1.0.19 - Adjusted the argument types of wmadgm and madgm. Added weighted distance wvdist. Improved the testing of vecvec.

  • Version 1.0.18 - Renamed madn to madgm (median of absolute deviations, i.e. radii, from gm). Added its weighted version wmadgm. They now take gm or wgm respectively as an argument, to avoid recomputation. Removed radvec, as it was a simple difference of gm and centroid.

  • Version 1.0.16 - Added tukeyvec and test of tukeyvec. Also changed usage of ran crate to its generic methods within vecvec test.

  • Version 1.0.14 - Some improvements of README.md.

  • Version 1.0.13 - Updated ran dev-dependency to "^0.3".

  • Version 1.0.12 - New random number generators are now in their own crate ran. It has been added here to development dependencies, where it properly belongs. tests.rs have been changed accordingly. No other changes.

  • Version 1.0.11 - The random number generators in indxvec have been been moved to their own module random. To keep tests.rs compatible, the import is now changed accordingly, to: use indxvec::random::*;

  • Version 1.0.10 Now using new random numbers generators from indxvec for testing.

  • Version 1.0.9 Removed genvec and genvecu8 as non-essential. Removed some examples in the code that were using them. Changed the printing of vecs to utilise the new trait Printing from indxvec. See testing.rs for usage.

  • Version 1.0.8 Pruned some non-essential code, such as smedian. Gmedian now performs consistently a bit better.

  • Version 1.0.7 Achieved further 20% speedup of gmedian by optimising some inner loops.

  • Version 1.0.6 Added crossfeatures - computes relationships between all pairs of vectors in self. Returns flattened lower triangular (symmetric) matrix.

    Dependence of two vectors is now normalised to the range [0,1], e.g. the dependence of two identical vectors without repetitions is 1. Same for vectors of any real values that are all unique. In these cases it is better to fall back to correlations. N-dependence of a whole set of vectors will often be more than one.

  • Version 1.0.5 Added 1D median correlation medaincorr. This is a more robust measure. Added dependencies and correlations which efficiently map these relationships of a single given vector (typically of outcomes), to a set of vectors (typically feature vectors).

  • Version 1.0.4 Added joint pdf, joint entropy and dependence for a set of n vectors.

  • Version 1.0.3 Better implementations of joint probability and joint entropy. Code style and testing improvements.

  • Version 1.0.2 Updated the dependency indxvec to version 1. A few minor changes to this document.

  • Version 1.0.1 Minor change: sortedeccs and wsortedeccs now take gm as an argument for more efficient repeated use. Vecvec test improved.

  • Version 1.0.0 Rstats reaches stability (of sorts)!