ozzie80 / go-miner

Data Mining Algorithms in Go

go-miner

Data Mining Algorithms in GoLang.

Installation

$ go get github.com/ozzie80/go-miner

Algorithms

k-means

Description From Wikipedia:

k-means clustering aims to partition n observations into k clusters in which each observation belongs to the cluster with the nearest mean, serving as a prototype of the cluster. This results in a partitioning of the data space into Voronoi cells.

The k-means implementation of go-miner uses the k-means++ algorithm for choosing the initial cluster centroids. The implementation provides internal quality indexes, Dunn Index and Davies-Bouldin Index, for evaluating clustering results.

Usage Example

// Create points or read from a .csv file
points := []dt.Vector{{1.0, 2.0, 3.0}, {5.1, 6.2, 7.3}, {2.0, 3.5, 5.0}}

// Specify Parameters: K, Points, MaxIter (optional) 
params := kmeans.Params{2, points, math.MaxInt64}

// Run k-means
clusters, err := kmeans.Run(params)

// Get quality index score
index := kmeans.DunnIndex{} // DaviesBouldinIndex{}
score := index.GetScore(clusters)

To Do

• Concurrent version
• Cuda/GPU version

To Be Added

• Apriori
• KNN
• Naive Bayes
• PageRank
• SVM
• ...

License

go-miner is MIT License.