Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/tpaschalis/xicor-go
A Go implementation of Chatterjee's Xi correlation coefficient
https://github.com/tpaschalis/xicor-go
go math statistics
Last synced: 9 days ago
JSON representation
A Go implementation of Chatterjee's Xi correlation coefficient
- Host: GitHub
- URL: https://github.com/tpaschalis/xicor-go
- Owner: tpaschalis
- License: mit
- Created: 2021-12-27T09:34:11.000Z (almost 3 years ago)
- Default Branch: main
- Last Pushed: 2022-01-03T14:56:51.000Z (almost 3 years ago)
- Last Synced: 2024-06-20T10:22:48.456Z (5 months ago)
- Topics: go, math, statistics
- Language: Go
- Homepage:
- Size: 40 KB
- Stars: 1
- Watchers: 3
- Forks: 0
- Open Issues: 3
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# xicor-go
[](https://opensource.org/licenses/MIT)
## What is xicor?
`xicor-go` is a Go implementation of the "xi" correlation metric described in the paper [_"A new coefficient of correlation"_](https://arxiv.org/pdf/1909.10140.pdf) by Sourav Chatterjee (arxiv.org/abs/1909.10140).
The current implementation is based off the [R code](https://statweb.stanford.edu/~souravc/xi.R) mentioned in the paper, and influenced by a [Python port](https://github.com/czbiohub/xicor) by [czbiohub](https://github.com/czbiohub/xicor). There is also a [mirror](https://github.com/cran/XICOR) of the CRAN R package hosted on GitHub.
The package provides the `Xi` struct which receives the input data along with a number of functional options and is used to calculate the correlation coefficient and p-value.
For a graphical comparison between the standard linear correlation coefficient and the Xi "rank correlation", you can refer to [this](https://twitter.com/adad8m/status/1474754752193830912) excellent tweet by [adad8m](https://twitter.com/adad8m)
## Installation
For a project utilizing Go modules, all you need to do is
```
go get github.com/tpaschalis/xicor-go
```
## Usage
```go
import "github.com/tpaschalis/xicor-go"
func main() {
x := []float64{0, 1, 2, 3, 4}
y := []float64{5, 6, 7, 8, 9}
// Create a new Xi object and get the correlation
xi, err := xicor.New(x, y).Correlation()
// Create a new Xi object and obtain the correlation along with its p-value
// Use functional options to define how to perform the calculation
xi, pvalue, err := xicor.New(
x,
y,
xicor.WithPermutationPvalue(1000),
xicor.WithoutTies(),
).Pvalue()
// You can also use the Xi object directly
data := &xicor.Xi{
X: x,
Y: y,
WantPvalue: true,
Method: "asymptotic",
DataTies: true,
}
xi, pvalue, err = data.Pvalue()
fmt.Println(xi, pvalue, err)
}
```
## Current status
I'm working towards a more stable and performant v0.0.1 release; the focus is on:
- Validating correctness of results by comparing against original R code (current tests haven't produced any inconsistency yet)
- Add support for categorical variables (a la the `factor` function used in R)
- Use dev branch to rewrite some of the code using Go 1.18 generics
- Run through a profiler to find and eliminate bottlenecks
## Benchmarks
The current pre-v0.0.1 version of the code runs through 100k randomized samples in about 0.12 seconds.
While there's a lot of low-hanging fruit to improve performance, I'll get to it once v0.0.1 is near.
## Contributing
If you have an idea, or would like to discuss and contribute an improvement, you can reach out in the repo [Issues](https://github.com/tpaschalis/xicor-go/issues) and open a [Pull Request](https://github.com/tpaschalis/xicor-go/pulls)
## License
This code in this repo is licensed under the [MIT license](LICENSE).