Random Swap

RandomSwap(
    local_search::AbstractAlgorithm
    verbose::Bool = DEFAULT_VERBOSE
    max_iterations::Integer = 200
    max_iterations_without_improvement::Integer = 150
)

RandomSwap is a meta-heuristic approach used for clustering problems. It follows an iterative process that combines local optimization with perturbation to explore the search space effectively. A local optimization algorithm is applied at each iteration to converge toward a local optimum. Then, a perturbation operator generates a new starting point and continues the search.

Fields

• local_search: the clustering algorithm applied to improve the solution in each meta-heuristics iteration.
• verbose: controls whether the algorithm should display additional information during execution.
• max_iterations: represents the maximum number of iterations the algorithm will perform before stopping.
• max_iterations_without_improvement: represents the maximum number of iterations allowed without improving the best solution.

References

• Fränti, Pasi. Efficiency of random swap clustering. Journal of big data 5.1 (2018): 1-29.

fit(
    meta::RandomSwap,
    data::AbstractMatrix{<:Real},
    k::Integer
)

The fit function applies a random swap to a clustering problem and returns a result object representing the clustering outcome.

Parameters:

• meta: an instance representing the clustering settings and parameters.
• data: a floating-point matrix, where each row represents a data point, and each column represents a feature.
• k: an integer representing the number of clusters.

Example

n = 100
d = 2
k = 2

data = rand(n, d)

kmeans = Kmeans()
random_swap = RandomSwap(local_search = kmeans)
result = fit(random_swap, data, k)