class FeedForward(number_of_individuals, num_parents_mating, death_percentage, number_of_generations, **kwargs)[source]#

A population of Feed Forward DNN topologies that will be evaluated utilizing a genetic search.

Parameters:

  • number_of_individuals (int) – The number of individuals that the user wants in the population.

  • num_parents_mating (int) – The number of parents that will be used to mate and create the next generation.

  • death_percentage (float) – The percentage of individuals that will die off each generation.

  • number_of_generations (int) – The number of generations that the population will be simulated for.

_evaluate()[source]#

Evaluate the fitness of all individuals in the population. :return:

_simulate()[source]#

Simulate the population for the number of generations. Callable method. :return:

fit(data, **kwargs)[source]#

Fit this population of FeedForward Individuals to the given data, and return the best model.

Parameters:

  • data – The data to fit this population to. Has to be either a pandas dataframe with columns: [feature1, feature2, …, featureN, label] or a file path to a file of formats specified in ref: dataLoaders.dataLoader.choose_data_loader that have the same layout.

  • **kwargs – Any additional arguments to pass to each FeedForwardIndividual which is a Keras model.

fit_predict(data, **kwargs)[source]#

Fit this population of FeedForward Individuals to the given data, and return the predictions of the best model.

Parameters:

data – The data to fit this population to. Has to be either a pandas dataframe with columns: [feature1, feature2, …, featureN, label] or a file path to a file of formats specified in ref: dataLoaders.dataLoader.choose_data_loader that have the same layout.

predict(X)[source]#

Predict the labels of the given data, utilizing the best found model.

Parameters:

X – The data to predict the labels of. Has to be either a pandas dataframe with columns: [feature1, feature2, …, featureN] or a file path to a file of formats specified in ref: dataLoaders.dataLoader.choose_data_loader that have the same layout.