amirdeljouyi / Genetic-Algorithm-on-K-Means-Clustering

Implementing Genetic Algorithm on K-Means and compare with K-Means++

Genetic Algorithm on K-Means Clustering

This Project is mainly based on the Genetic-Kmeans-Algorithm-GKA-

The approaches which I used

• Min-max normalization for standardization
• Davies–Bouldin index for evaluation of each cluster
• IN GENETIC :
  • Rank-based selection
  • One-point crossover

Requirements

• Panda
• NumPy

Getting Started

python __main__.py

Input

• The data that I analyzed is from Iris
  • data/iris.csv have 3 column and data/iris2.csv have 4 column and data/isis_with_header.csv with header
• config.txt contain control parameters
  • kmax: maximum number of clusters
  • budget: budget of how many times run GA
  • numOInd: number of Individual
  • Ps: the probability of ranking Selection
  • Pc: the probability of crossover
  • Pm: the probability of mutation

Output

• norm_data.csv is normalization data
• cluster_json is centroid of each cluster
• result.csv is data with labeled to each cluster

Analysis

• the accuracy of GA on K-means: 88%
• the accuracy of k-means++: 83%

Reference

• Genetic-Kmeans-Algorithm-GKA-