def main():
target_string = "Genetic Algorithm"
population_size = 100
mutation_rate = 0.05
genetic_algorithm = GeneticAlgorithm(target_string,
population_size,
mutation_rate)
print ("")
print ("+--------+")
print ("| GA |")
print ("+--------+")
print ("Description: Implementation of a Genetic Algorithm which aims to produce")
print ("the user specified target string. This implementation calculates each")
print ("candidate's fitness based on the alphabetical distance between the candidate")
print ("and the target. A candidate is selected as a parent with probabilities proportional")
print ("to the candidate's fitness. Reproduction is implemented as a single-point")
print ("crossover between pairs of parents. Mutation is done by randomly assigning")
print ("new characters with uniform probability.")
print ("")
print ("Parameters")
print ("----------")
print ("Target String: '%s'" % target_string)
print ("Population Size: %d" % population_size)
print ("Mutation Rate: %s" % mutation_rate)
print ("")
genetic_algorithm.run(iterations=1000)
def greedy_evolver_test():
from angular_evolver import edge_order_evolver as OE
crossover = OE.OrderUniformCrossover()
selection = OE.uniform_wheel_selection
mutation = OE.EdgeOrderMutation()
solver = OE.AngularMinSumGreedySolver()
result_sol_func = lambda x: np.array([item.solution[1] for item in x])
fitness = OE.AngularSolverFitness(
solver.__class__.__name__,
remote_computation=False,
fitness_goal=OE.AngularSolverFitness.MIN_SUM,
solver_params={"no_sol_return": True},
custom_result_func=result_sol_func)
termination = OE.IterationTerminationConditionMet(max_iter=200)
callback = OE.update_callback
graph = OE.create_circle_n_k(12, 3)
init_pop = np.zeros(200, dtype=object)
init_pop[:] = [OE.EdgeOrderGraphGenome(graph) for i in range(200)]
from genetic_algorithm import GeneticAlgorithm
ga = GeneticAlgorithm(genomes=init_pop,
selection=selection,
mutation=mutation,
fitness=fitness,
crossover=crossover,
termCon=termination,
callback=callback)
last = ga.evolve()
return last
def lambda_handler(list_elites_schedules, param):
# Recupere tous les emplois du temps des fonctions Lambda precedentes
all_elites = []
for schedules_list in list_elites_schedules:
for schedule in schedules_list:
all_elites.append(schedule)
# Tri l'ensemble des emplois du temps par ordre decroissant de fitness
sorted_list = sorted(all_elites,
key=lambda schedule: schedule._fitness,
reverse=True)
# Creation de la population avec les elites des populations precedentes
data = Data()
newPop = Population(size=param[0][0], data=data, schedules=sorted_list)
algo = GeneticAlgorithm(data=data, param=param[0])
# Evolution de la population en local sur les serveurs d'AWS
for _ in range(param[0][6]): # param[0][6]==NUMB_OF_GENERATION
if (newPop.schedules[0]._fitness != 1.0):
newPop = algo.evolve(population=newPop)
for schedule in newPop.schedules:
schedule._fitness = schedule.calculate_fitness()
newPop.sort_by_fitness()
# Recupere les elites pour les retourner
elites = []
for i in range(param[0][5]): # param[0][5]==NUMB_OF_ELITES
elites.append(newPop.schedules[i])
return elites
def main():
iris = datasets.load_iris()
X = iris.data
y = iris.target
n_clusters = 3
pop_size = 100
mutation_rate = 0.01
crossover_rate = 0.85
elitism_count = 3
ga = GeneticAlgorithm(n_clusters, pop_size, mutation_rate, X,
crossover_rate, elitism_count)
d1, y1 = datasets.make_blobs(n_samples=121, centers=3, n_features=5)
d2, y2 = datasets.make_blobs([378, 233], 2)
d3, y3 = datasets.make_blobs([220, 145, 56, 378, 117], 5)
d4, y4 = datasets.make_blobs(100, 8)
ga1 = GeneticAlgorithm(n_clusters, pop_size, mutation_rate, d1,
crossover_rate, elitism_count)
ga2 = GeneticAlgorithm(n_clusters, pop_size, mutation_rate, d2,
crossover_rate, elitism_count)
ga3 = GeneticAlgorithm(n_clusters, pop_size, mutation_rate, d3,
crossover_rate, elitism_count)
ga4 = GeneticAlgorithm(n_clusters, pop_size, mutation_rate, d4,
crossover_rate, elitism_count)
run_algo(ga, X, pop_size)
run_algo(ga1, d1, pop_size)
run_algo(ga2, d2, pop_size)
run_algo(ga3, d3, pop_size)
run_algo(ga4, d4, pop_size)
def main():
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
tokenize = MosesTokenizer('en')
detokenize = MosesDetokenizer('en')
def print_output(output):
for _r in output:
logger.info("%s: %s", detokenize(_r), str(_r.fitness.values))
logger.info(" ")
population_size = 25
max_iter = 5
crossover_prob = 0.2
mutation_prob = 0.8
ga_weights = weights
generator = GeneticAlgorithm(population_size=population_size,
max_iter=max_iter,
crossover_prob=crossover_prob,
mutation_prob=mutation_prob,
weights=ga_weights)
text = 'Suffering from Success'
tokens = tokenize(text)
pop, _, hof = generator.run(tokens)
print_output(hof)
tokenize.close()
detokenize.close()
def process_task(config, task, session):
# First init all callable classes
try:
mutation = _instantiate_callables(config.mutation_func, None)
selection = _instantiate_callables(config.selection_func, None)
crossover = _instantiate_callables(config.crossover_func, None)
fitness = _instantiate_callables(config.fitness_func,
config.fitness_func_initargs)
if config.term_condition == 'IterationTerminationConditionMet' and not config.term_condition_initargs:
term_con = IterationTerminationConditionMet(
max_iter=config.generations)
else:
term_con = _instantiate_callables(config.term_condition,
config.term_condition_initargs)
if config.callback == 'SaveCallback' and config.callback_initargs is None:
callback = SaveCallback(config.generations,
config.population_amount, task, session)
else:
callback = _instantiate_callables(config.callback,
config.callback_initargs)
task.status = Task.STATUS_OPTIONS.PROCESSING
if session:
session.commit()
# Now load population if provided, else generate it
starting_generation, population = _load_population(
config, task, session)
if config.PreEvolveInteractive:
print(
"Config set up. To change the population just change the 'population' variable."
)
print("For other variables just refer to the locals.")
embed()
gen_algo = GeneticAlgorithm(
genomes=population,
selection=selection,
mutation=mutation,
fitness=fitness,
crossover=crossover,
callback=callback,
termCon=term_con,
elitism=config.elitism,
mutationChance=config.mutation_chance_genome,
mutationChanceGene=config.mutation_chance_gene)
gen_algo.evolve(generation=starting_generation)
task.status = Task.STATUS_OPTIONS.FINISHED
if session:
session.commit()
except InterruptedError as e:
task.status = task.STATUS_OPTIONS.INTERRUPTED
if session:
session.commit()
except Exception as e:
if session:
task.status = Task.STATUS_OPTIONS.ERROR
task.error_message = str(e)
session.commit()
print(e)
raise e
def run():
global SCHEDULE_NUMBER, CLASS_NO
from data import Data
from genetic_algorithm import GeneticAlgorithm
from population import Population
generation_number = 0
data = Data()
_genetic_algorithm = GeneticAlgorithm(data=data)
_population = Population(size=POPULATION_SIZE, data=data).sort_by_fitness()
print_data(data=data)
print
print
print_population_schedules(population=_population,
generation_number=generation_number)
print_schedule_as_table(data=data,
schedule=_population.schedules[0],
generation=generation_number)
while _population.schedules[0].fitness != 1.0:
generation_number += 1
_population = _genetic_algorithm.evolve(
population=_population).sort_by_fitness()
print_population_schedules(population=_population,
generation_number=generation_number)
print_schedule_as_table(data=data,
schedule=_population.schedules[0],
generation=generation_number)
def execDir(self):
global pathh
pat =pathh
files = listdir(pathh)
open("result_ag.csv", "w").close()
print(1)
print(pat)
for f in files:
l = pat + "/" + f
print(f)
j = self.InstanceReader2(f)
print("hada",j)
instance, n, C, w = self.InstanceReader(l)
items = [Item(size=int(i)) for i in w]
shuffle(items)
thing = GeneticAlgorithm(C, items)
# c, conf, time = self.AG(weight, C)
start = timeit.default_timer()
thing.run()
print(6)
stop = timeit.default_timer()
time = stop - start
c = thing.best_solution.num_bins
print(9)
resultatt = open("result_ag.csv", mode="a+")
resultatt.write(f'{instance},{n},{C},{c},{j},{time * 1000}\n') # Ecrire les resultats dans un fichier csv
resultatt.close()
print(0)
self.loadData_mod_csv('./result_ag.csv')
def setUp(self):
# initializing variables for self.ga
self.expected_input_size = 2
self.expected_hidden_size = 5
self.expected_output_size = 1
self.population_size = 10
self.selection_size = 2
self.learning_rate = 1e-3
self.epochs = 5
self.generations = 2
self.ga = GeneticAlgorithm(
False,
self.expected_input_size,
self.expected_hidden_size,
self.expected_output_size,
self.population_size,
self.selection_size,
self.learning_rate,
self.epochs,
self.generations,
)
self.hga = GeneticAlgorithm(
True,
self.expected_input_size,
self.expected_hidden_size,
self.expected_output_size,
self.population_size,
self.selection_size,
self.learning_rate,
self.epochs,
self.generations,
)
pass
def run(file, seed, population_size, mutation_rate, elitism_rate, output_file):
instance = Instance.load(file)
algorithm = GeneticAlgorithm(instance,
seed,
population_size,
mutation_rate,
elitism_rate)
start_time = time.time()
first_solution, solution = algorithm.run()
stop_time = time.time()
print(f'Solution fitness: {solution[1]}')
print(f'Execution time: {stop_time - start_time}')
solution_edges = get_solution_edges(solution)
output = {
'parameters': {
'instance': file,
'seed': seed,
'population-size': population_size,
'mutation-rate': mutation_rate,
'elitism-rate': elitism_rate
},
'first-solution-fitness': first_solution[1],
'solution-fitness': solution[1],
'execution-time': stop_time - start_time,
'solution': solution_edges
}
with open(output_file, mode='w') as out_file:
out_file.write(json.dumps(output, indent=2))
def run_algorithm(data, labels, output_file, learner):
"""Runs the genetic algorithm"""
fitness_calc = None
if learner == 'SVM':
fitness_calc = SVCFitnessCalc(data, labels)
elif learner == 'DT':
fitness_calc = DTFitnessCalc(data, labels)
elif learner == 'MLP':
fitness_calc = MLPFitnessCalc(data, labels)
else:
print 'Learner error'
ga = GeneticAlgorithm(fitness_calc, learner)
with open(output_file, 'a') as csvfile:
out_writer = csv.writer(csvfile)
out_writer.writerow(['time', 'generation', 'parameters', 'fitness'])
try:
start_time = time.time()
pop = Population(size=50,
fit_calc=fitness_calc,
initialize=True,
indiv_type=learner)
generation = 1
fittest = pop.get_fittest()
fitness = fittest.get_fitness()
best_fitness = fitness
params = fittest.get_params_str()
cur_time = time.time() - start_time
with open(output_file, 'a') as csvfile:
out_writer = csv.writer(csvfile)
out_writer.writerow([cur_time, generation, params, best_fitness])
while fitness < fitness_calc.max_fitness:
pop = ga.evolve_pop(pop)
generation += 1
fittest = pop.get_fittest()
fitness = fittest.get_fitness()
if fitness > best_fitness:
best_fitness = fitness
params = fittest.get_params_str()
cur_time = time.time() - start_time
with open(output_file, 'a') as csvfile:
out_writer = csv.writer(csvfile)
out_writer.writerow(
[cur_time, generation, params, best_fitness])
sys.exit()
except ZeroDivisionError, e:
with open(output_file, 'a') as csvfile:
out_writer = csv.writer(csvfile)
out_writer.writerow(['ZeroDivisionError'])
out_writer.writerow([e])
sys.exit()
def run(self):
result1 = GeneticAlgorithm(self.bin_capacity, self.items,self.POPULATION_SIZE,self.MAX_GENERATIONS,self.MAX_NO_CHANGE,self.TOURNAMENT_SIZE,self.MUTATION_RATE,self.CROSSOVER_RATE)
total_iterationsAG, stagnationAG = result1.run()
result2 = TabuSearch(self.bin_capacity, self.items,self.MAX_COMBINATION_LENGTH,self.MAX_ITERATIONS,self.MAX_NO_CHANGE2)
total_iterationsTB, stagnationTB, combinationTB = result2.run2(result1)
return result1,result2,total_iterationsAG, stagnationAG, total_iterationsTB, stagnationTB, combinationTB
def run(self):
result1 = GeneticAlgorithm(self.bin_capacity, self.items, self.POPULATION_SIZE,self.MAX_GENERATIONS,self.MAX_NO_CHANGE,self.TOURNAMENT_SIZE,self.MUTATION_RATE,self.CROSSOVER_RATE)
total_iterationsAG, stagnationAG = result1.run()
sa = SA(self.alpha,self.bin_capacity, self.items,self.t_init, self.t_target, self.iter_nb)
sa.run_for_hrh(result1.best_solution.generate_solution(self.items))
return sa
def main():
utils.delete_files()
data = Data()
if data.algorithm == Algorithm.GENETIC:
algorithm = GeneticAlgorithm(data)
else:
algorithm = CSPAlgorithm(data)
algorithm.run()
print("Finished!")
def single_test(steps, population_size, crossover_chance, mutation_chance, evaluate_number):
sudoku = get_initial_sudoku()
ga = GeneticAlgorithm(sudoku)
solver = GeneticSolver(
sudoku,
steps = steps,
population_size = population_size,
crossover_chance = crossover_chance,
mutation_chance = mutation_chance
)
population = solver.solve()
population.sort(key=lambda ind: ga.fitness_function(ind))
ind = ga.get_individual_from_indexed_list(population[-1:][0])
sudoku.fill_missing(ind)
print(sudoku)
evaluation = sudoku.evaluate()
print(evaluation)
evaluate_number = evaluate_number + sudoku.evaluate()
um, dois, tres, quatro, cinco, seis, sete, oito, nove = 0,0,0,0,0,0,0,0,0
for i in sudoku.puzzle:
if i == 1:
um += 1
elif i == 2:
dois += 1
elif i == 3:
tres += 1
elif i == 4:
quatro += 1
elif i == 5:
cinco += 1
elif i == 6:
seis += 1
elif i == 7:
sete += 1
elif i == 8:
oito += 1
elif i == 9:
nove += 1
print('\nUm', um)
print('Dois', dois)
print('Tres', tres)
print('Quatro', quatro)
print('Cinco', cinco)
print('Seis', seis)
print('Sete', sete)
print('Oito', oito)
print('Nove', nove)
return evaluate_number
def display(self):
pygame.init()
pygame.display.set_caption('MLCars2D')
screen = pygame.display.set_mode(self.size)
icon = pygame.image.load('resources/images/icon.png')
pygame.display.set_icon(icon)
self.game_objects_controller = GameObjectsController(
self.window_width, self.window_height, screen, self)
self.game_objects_controller.display_menu()
self.game_objects_controller.number_of_cars = constants.CARS_NO
self.game_objects_controller.initialize_track_with_random_cars()
keyboardEvents = KeyboardEventHandler()
iteration_number = 0
self.genetic_algorithm = GeneticAlgorithm(
38, self.game_objects_controller.number_of_cars, 40, 0.05)
population = self.genetic_algorithm.perform_crossing_and_mutation()
print(
len(self.game_objects_controller.cars[0].neural_network.
get_weight_list()))
self.game_objects_controller.reinitialize_cars(population)
clock = pygame.time.Clock()
fps = constants.FPS
simulation_length = 400
while True:
if self.game_objects_controller.play_action_frame_count % simulation_length == 0:
distances = self.game_objects_controller.get_car_distances()
population = self.perform_learning_iteration(
distances, population, self.genetic_algorithm,
iteration_number)
print(sum(distances) / (len(distances)))
self.game_objects_controller.update_stat_box()
self.game_objects_controller.reinitialize_cars(population)
keyboardEvents.reset()
for event in pygame.event.get():
if event.type == pygame.QUIT:
sys.exit()
if event.type == pygame.MOUSEBUTTONUP:
pass
keyboardEvents.process_events(event)
self.game_objects_controller.check_pressed_buttons(event)
self.game_objects_controller.perform_action(keyboardEvents)
pygame.display.flip()
clock.tick(fps)
def genetic_iterations(graph,ite):
nTournament = 6
probability = 0.05
nPopulation = 50
nIterations = ite
nElite = 10
genetic = GeneticAlgorithm(
nPopulation, nIterations, nElite, nTournament, probability, graph)
solution = genetic.genetic_algorithm()
return solution.fitness
def demonstrate_genetic_algorithm(pop_size, elite_size, rate, gens):
print("\t\tGenetic Algorithm")
print("Initial Population Size: {}".format(pop_size))
print("Elite Size: {}".format(elite_size))
print("Mutation Rate: {}".format(rate))
print("Generations: {}".format(gens))
ga = GeneticAlgorithm(graph, pop_size, elite_size, rate, gens)
tour, t = ga.run()
print("Shortest Tour: {}".format(tour[0]))
print("Cost: {}".format(tour[1]))
print("Number of Seconds Taken: {}\n".format(t))
def create_heatmap(problem_obj):
"""
Creates de HeatMap.
:param problem_obj: a BitSequence, Phrase or Knapsack instance
"""
limit_generation = 500
mutation_rate = np.array([0, 0.1, 0.2, 0.3, 0.4, 0.5])
population_size = np.arange(100, 750, 50)
heatmap = np.zeros((mutation_rate.shape[0], population_size.shape[0]))
for rate, i in zip(mutation_rate, range(mutation_rate.shape[0])):
for size, j in zip(population_size, range(population_size.shape[0])):
# Initialize the genetic algorithm
genetic_algorithm = GeneticAlgorithm(
population_size=size,
fitness=problem_obj.fitness,
generate_gene=problem_obj.generate_gene,
generate_individual=problem_obj.generate_individual,
mutation_rate=rate,
condition_finish=problem_obj.condition_finish,
max_generation=limit_generation)
# Calculates how many generations the algorithm takes
generations_data = genetic_algorithm.run_algorithm(verbose=False)
# Puts the result in the HeatMap
heatmap[i, j] = generations_data.shape[0]
# Creates HeatMap
sns.set()
plt.figure(figsize=(10, 8))
fig, ax = plt.subplots()
sns.heatmap(heatmap,
annot=True,
annot_kws={'va': 'top'},
ax=ax,
vmin=0,
vmax=limit_generation,
fmt=".1f",
linewidths=0.25,
square=True,
cbar=False,
xticklabels=population_size,
yticklabels=mutation_rate)
plt.title('Population size v/s Mutation rate v/s Generation')
plt.xlabel('Population size')
plt.ylabel('Mutation rate')
plt.show()
def genetic_population(graph,pop):
nTournament = 6
probability = 0.05
nPopulation = pop
nIterations = 500
nElite = 30
genetic = GeneticAlgorithm(
nPopulation, nIterations, nElite, nTournament, probability, graph)
solution = genetic.genetic_algorithm()
return solution.fitness
def __init__(self, algo, calib_type, hp_grid, cv, scoring, n_iter=None):
self.cv = cv
self.algo = algo
self.scoring = scoring
self.hp_grid = hp_grid
self.best_hp = {}
if calib_type == 'GridSearch':
self.hp_iterator = ParameterGrid(self.hp_grid)
elif calib_type == 'RandomSearch':
self.hp_iterator = ParameterSampler(self.hp_grid, n_iter)
elif calib_type == 'GeneticAlgorithm':
self.hp_iterator = GeneticAlgorithm(self.hp_grid, n_iter)
def __init__(self,
inputs=7,
outputs=3,
hidden_layers=[],
population_size=20,
fittest_percent=0.2,
mutation_chance=0.05,
crossover_points=1,
screen_size=20,
delay=200,
box_width=20,
food_value=200,
moves_to_decrement=1,
score_decrement=2,
score_increment=2,
score_decrement_move=2,
turn_decrement_factor=1.5,
initial_score=100,
score_tracking=False,
save_to_file=False):
self.population_size = population_size
self.generation = 0
self.current_individual_index = 0
self.decrement_factor = turn_decrement_factor
self.snake_list = []
self.network_list = []
self.fitness_list = np.zeros(self.population_size)
# Recording best snake
self.max_fitness = -np.inf
self.max_fitness_network = None
self.save_to_file = save_to_file
self.algorithm = GeneticAlgorithm(fittest_percent, mutation_chance,
crossover_points)
for i in range(self.population_size):
self.network_list.append(
NeuralNetwork(inputs, outputs, hidden_layers))
super().__init__(screen_size=screen_size,
delay=delay,
box_width=box_width,
human_player=False,
food_value=food_value,
moves_to_decrement=moves_to_decrement,
score_decrement=score_decrement,
score_increment=score_increment,
initial_score=initial_score,
score_tracking=score_tracking,
score_decrement_move=score_decrement_move,
turn_decrement_factor=turn_decrement_factor)
def natural_selection(dataset):
ga = GeneticAlgorithm()
neural_net_population = ga.create_initial_population()
for i in range(Constants.num_generations - 1):
print("Generation {}".format(i), '-' * 20)
train_population(neural_net_population, dataset)
avg_fitness = calculate_avg_fitness(neural_net_population)
print('*' * 5, "Generation {}: {}".format(i, avg_fitness))
neural_net_population = ga.evolution(neural_net_population)
show_results(neural_net_population)
def run_genetic_algorithm(seconds, population_size, crossover, mutate):
print('Running Genetic Algorithm for ' + str(seconds) + ' seconds...')
print()
ga = GeneticAlgorithm(states, seconds, population_size, crossover, mutate,
inc_support, dec_support)
ga.run()
print('Found optimal route with value of ' + str(ga.best_solution.value) +
'.')
print(
str(ga.best_solution.calculate_real_value()) +
' electoral votes were collected.')
ga.best_solution.print()
print()
def _run_trials(parameters: Parameters, trials: int):
results = np.empty(shape=(trials, ), dtype=bool)
ga = GeneticAlgorithm(parameters)
start_time = time.time()
for i in range(trials):
results[i] = ga.run()
finish_time = time.time()
average_time = (finish_time - start_time) / trials
success_rate = np.count_nonzero(results) / trials
return (average_time, success_rate)
def main(argv):
# Variables for genetic algorithm
number_of_generations = 30
population_size = 100
crossover_probability = 0.7
mutation_probability = 0.1
# Variables for neutral network
number_of_hidden_layers = 3
node_for_hidden_layer = 6
# Variables for snake game
draw = False
board_size = 10
log = False
try:
opts, args = getopt.getopt(argv, "hdLg:p:c:m:l:n:b:",
["generations=", "population=", "crossover=", "mutation=", "layers=",
"nodes=", "board="])
except getopt.GetoptError:
print(
'main.py -d (draw best run) -L (save every chromosome with its score to log file) -g <number_of_generations> -p <population_size> -c <crossover_probability>\
-m <mutation_probability> -l <number_of_hidden_layers> -n <nodes_for_hidden_layer> -b <board_size>')
sys.exit(2)
for opt, arg in opts:
if opt == '-h':
print(
'main.py -d (draw best run) -L (save every chromosome with its score to log file) -g <number_of_generations> -p <population_size> -c <crossover_probability>\
-m <mutation_probability> -l <number_of_hidden_layers> -n <nodes_for_hidden_layer> -b <board_size>')
sys.exit()
elif opt == '-d':
draw = True
elif opt == '-L':
log = True
elif opt in ("-g", "--generations"):
number_of_generations = int(arg)
elif opt in ("-p", "--population"):
population_size = int(arg)
elif opt in ("-c", "--crossover"):
crossover_probability = float(arg)
elif opt in ("-m", "--mutation"):
mutation_probability = float(arg)
elif opt in ("-l", "--layers"):
number_of_hidden_layers = int(arg)
elif opt in ("-n", "--nodes"):
node_for_hidden_layer = int(arg)
elif opt in ("-b", "--board"):
board_size = int(arg)
print(number_of_generations, population_size, crossover_probability, mutation_probability, number_of_hidden_layers,
node_for_hidden_layer, board_size)
genetic_algorithm = GeneticAlgorithm(population_size, number_of_generations, crossover_probability,
mutation_probability, board_size, draw, log)
genetic_algorithm.number_of_hidden_layers = number_of_hidden_layers
genetic_algorithm.node_for_hidden_layer = node_for_hidden_layer
genetic_algorithm.init_population()
genetic_algorithm.evolve()
def main_show_maze(self, pop, cross, mut, max_generation, elitismo):
# vars
solution = False
generation = 0
mut = float(mut)
cross = float(cross)
elitismo = True if str(elitismo) == '1' else False
# Cria a população
##########################################3
population = Population(tamPop=int(pop))
population.create_random_population()
# Cria o algoritmo genetico
algoritmo = GeneticAlgorithm(crossover=cross, mutation=mut, elitismo=elitismo)
while solution == False and generation<=int(max_generation):
generation += 1
population = algoritmo.create_new_generation(current_population=population)
best_specimen = population.getSpeciemen(0)
# Reseta a tela e mostra o caminho do melhor cromossomo
######################################################################
if generation % 10 == 0:
self.env.reset()
self.env.after(100, self.show_path_on_maze(best_specimen=best_specimen))
solution = algoritmo.get_solution(best_specimen.genetic_code)
conteudo = '\nCódigo={} \nFitness={} \nGeração={}\nRota=[{}]'.format(best_specimen.genetic_code, best_specimen.fitness, generation,best_specimen.route)
self.env.delete()
self.env.escreve(conteudo)
print(best_specimen.genetic_code, best_specimen.fitness, best_specimen.route, generation)
if solution == True:
conteudo = 'Solução ótima!!' \
'\nCódigo={} ' \
'\nFitness={} ' \
'\nGeração={}'.format(best_specimen.genetic_code,
best_specimen.fitness, generation)
self.env.delete()
self.env.escreve(conteudo)
print(best_specimen.genetic_code, best_specimen.fitness, best_specimen.route, generation)
i = 0
while i <= 10:
i+=1
self.env.reset()
self.env.after(100, self.show_path_on_maze(best_specimen=best_specimen))
class CrossVal():
""" Cross Validation general class for GridSearch, RandomSearch, and GeneticAlgorithm
GridSearch
Performs an Exhaustive Search for optimal hyperparameters inside a cross validation process
RandomSearch
Performs a Random Search for optimal hyperparameters inside a cross validation process
GeneticAlgorithm
NOT CODED YET
"""
def __init__(self, algo, calib_type, hp_grid, cv, scoring, n_iter=None):
self.cv = cv
self.algo = algo
self.scoring = scoring
self.hp_grid = hp_grid
self.best_hp = {}
if calib_type == 'GridSearch':
self.hp_iterator = ParameterGrid(self.hp_grid)
elif calib_type == 'RandomSearch':
self.hp_iterator = ParameterSampler(self.hp_grid, n_iter)
elif calib_type == 'GeneticAlgorithm':
self.hp_iterator = GeneticAlgorithm(self.hp_grid, n_iter)
def compute_cv(self, X, Y):
""" Function that operate the cross validation
Let us notive that this code is similar for the three different cross validation
However the iterator that will yield the tested parameters differs
"""
best_score = -np.Inf
for hp in self.hp_iterator:
self.algo.set_hyperparams(**hp)
score = []
for train, test in self.cv.split(X, Y):
self.algo.fit(X.iloc[train], Y.iloc[train])
pred_values = self.algo.predict(
X.iloc[test]
) # Be careful not to stock predicted values in the algo, since it is only temporary internal results
self.algo.compute_outputs(Y.iloc[test], pred_values,
self.scoring)
score.append(getattr(self.algo, self.scoring))
self.algo.reset_outputs(
) # For safety, it is currently needed, please do not change without rethinking the code
score_mean = np.mean(score)
if isinstance(self.hp_iterator,
GeneticAlgorithm) == 'GenericAlgorithm':
self.hp_iterator.update_score(score_mean)
if score_mean > best_score:
best_score = score_mean
self.best_hp = hp
return self
def run_genetic_algorithm_parallel(self, i):
validated_data = copy.copy(self.data[list(
self.valid_attributes.columns)])
validated_data[self.response] = self.response_data.values
ga = GeneticAlgorithm(data=validated_data,
response=self.response,
attributes=list(self.valid_attributes.columns),
evaluation=self.evaluation,
seed=i,
one_out=self.one_out,
model_config=self.model_config,
**self.ga_config["config"])
ga.execute(**self.ga_config["mutate_config"])
return ga.best_models.s_heap
def __init__(self, m=25.0, c=25.0, RedStd=5.0, randomseed=250192):
"""
Instantiates the class by synthetically generating data.
"""
GeneticAlgorithm.__init__(self,
nParams=2,
nPop=100,
nGen=1000,
p_m=0.05)
self.X = np.linspace(-10, 10, 25)
self.delta = np.random.uniform(low=-1 * RedStd,
high=RedStd,
size=len(self.X))
self.Y = (m * self.X + c) + self.delta
def genetic(graph):
cached.clear()
start = time.time()
nTournament = 6
probability = 0.05
nPopulation = 100
nIterations = 500
nElite = 30
genetic = GeneticAlgorithm(
nPopulation, nIterations, nElite, nTournament, probability, graph)
solution = genetic.genetic_algorithm()
end = time.time()
duration = end - start
return duration
from genetic_algorithm import GeneticAlgorithm GA = GeneticAlgorithm() GA.run()
