evo_spotis.stochastic_algorithms.de

Module Contents

Classes

DE_algorithm

class evo_spotis.stochastic_algorithms.de.DE_algorithm(var_min=sys.float_info.epsilon, var_max=1.0, max_it=200, n_pop=60, beta_min=0.2, beta_max=0.8, p_CR=0.4)[source]
__call__(X_train, y_train, types, bounds, verbose=True)[source]

Determine criteria weights using DE algorithm with the goal (fitness) function using SPOTIS method and Spearman rank coefficient

Parameters

  • X_train (ndarray) – Decision matrix containing training dataset of alternatives and their performances corresponding to the criteria

  • y_train (ndarray) – Ranking of training decision matrix which is the targer variable

  • types (ndarray) – Criteria types. Profit criteria are represented by 1 and cost by -1.

  • bounds (ndarray) – Bounds contain minimum and maximum values of each criterion. Minimum and maximum cannot be the same.

  • verbose (bool) – For True verbose value, which is default, information about Best Fitness value in each iteration will be displayed and for False value, it will not

Returns

  • ndarray – Values of best solution representing criteria weights

  • ndarray – Best values of fitness function in each iteration required for visualization of fitness function.

  • ndarray – Mean values of fitness function in each iteration required for visualization of fitness function.

Examples

>>> de_algorithm = DE_algorithm()
>>> weights, BestFitness, MeanFitness = de_algorithm(X_train, y_train, types, bounds)
fitness_function(matrix, weights, types, bounds, y_train)[source]
_generate_population(X_train, y_train, types, bounds)[source]
_crossover(u, v, aj)[source]
static _de_algorithm(self, X_train, y_train, types, bounds)[source]