Skip to content

Latest commit

 

History

History
62 lines (47 loc) · 1.88 KB

README.md

File metadata and controls

62 lines (47 loc) · 1.88 KB
Raw

Build Status

gostats

Statistics for golang

Usage

To install, do go get github.com/r0fls/gostats. Check out gostats_test.go for a working example of using each distribution.

Example

package main

import (
    "fmt"
    "github.com/r0fls/gostats"
)

func main() {
    b := stats.Bernoulli(.5)
    fmt.Println(b.Random())

    // or fit a distribution from data...
    b = stats.FitBernoulli([0,1,1,1])
    fmt.Println(b.Random())
}

Distributions

Currently the following distributions are included:

  • Bernoulli
  • Laplace
  • Poisson
  • Geometric
  • Weibull
  • Exponential
  • Binomial
  • NegativeBinomial *

*note the negative binomial takes parameters r, p where r is the number allowed success before termination of the trials and p is the success of a given trial. The random variable produced by NegativeBinomial is the number of failures.

And each distribution has these functions:

  • Pmf or Pdf
  • Cdf
  • Quantile
  • Random

Also there is a corresponding function named FitDistrbution for each distribution, as shown in the above example with the Bernoulli. That function uses the MLE for each distribution to choose the best estimation for the parameters and returns an initialized distribution with them.

Common

Additionally there are some common functions. Most notably is LSR, which performs a least squares regression.

Additional Info

The default seed function is time.Now().UTC().UnixNano(). However you can override that by calling rand.Seed(uniqueInteger) yourself before generating random numbers.

Contributing

If you're interested in contributing, please submit a pull request, or raise an issue.

TO-DO
  • add more distributions
  • should random return an array?
  • allow updating a fitted distribution with more data