deleomike / sci-go

The Go Data Science Library

Sci-Go

The Go Data Science Library

This library is a data science library, not necessarily a Machine Learning or Deep Learning library. SciGo has been developed with an idiomatic design process, and inspiration from a few sources.

• Go Proverbs With Rob Pike

Why Should I Use SciGo?

1. SciGo doesn't lock you in
   • Inputs are almost always native Golang types, allowing you to interface with any awesome golang library like GoDataframe TODO ref
   • We rely on CGo as little as possible (TODO, ref on how CGo causes lots of problems), yet have good performance (TODO, back this up)
2. SciGo is very easy to use
   • SciGo has a very clear and organized design pattern
   • We painstakingly limit our dependencies to make your life easier.
3. SciGo is ambitious and has a solid roadmap for a ton of features

Installing

go get -u github.com/deleomike/sci-go

Usage

There are a lot of things in the works, but right now we have these feature sets implemented.

• Clustering
  • KMeans
  • Mean Shift

Support the Project

I appreciate all contributions and sponsorships for this project.

sci-go's Issues

Pretty Project Description

Gaussian Mixture

Kmeans

DBSCAN

MeanShift

Spectral Clustering

Implement Data Science Algorithms - WIP

Classifiers

• Decision Tree
• Random Forrest
• Perceptron

Datasets

• Make Blobs

Tree Structures

• KDTree

Preprocessing

• MinMaxScaler
• StandardScaler

Dimensionality Reduction

• PCA

Polynomial

• Interpolation
• Divided Difference

Clustering

• #2
• #6
• #7
• #8
• #9
• #10
• #5
• #11
• #12

Elbow Method

Benchmarks

BIRCH

OPTICS

Agglomerative Clustering