def compare_all(waldo_df, experiment_settings=DEFAULT_EXPERIMENT_SETTINGS):
opt_problem = WaldoOpt(waldo_df)
domain = opt_problem.domain
start = time.time()
rhc = hillclimb(domain=domain,
costf=opt_problem.compute_fitness,
max_evaluations=experiment_settings['rhc']['evaluations'])
rhc['time'] = time.time() - start
rhc.set_index('evaluations', inplace=True)
start = time.time()
sa = simulated_annealing(domain=domain,
costf=opt_problem.compute_fitness,
T=experiment_settings['sa']['T'])
sa['time'] = time.time() - start
sa.set_index('temperature', inplace=True)
start = time.time()
ga = genetic_optimize(domain=domain,
costf=opt_problem.compute_fitness,
maxiter=experiment_settings['ga']['generations'])
ga['time'] = time.time() - start
ga.set_index('generations', inplace=True)
start = time.time()
mm = mimic.run_mimic(domain=domain,
fitness_function=opt_problem.compute_fitness,
evaluations=experiment_settings['mm']['evaluations'])
mm['time'] = time.time() - start
return rhc, sa, ga, mm
def compare_all(waldo_df, experiment_settings=DEFAULT_EXPERIMENT_SETTINGS):
opt_problem = WaldoOpt(waldo_df)
domain = opt_problem.domain
start = time.time()
rhc = hillclimb(domain=domain,
costf=opt_problem.compute_fitness,
max_evaluations=experiment_settings['rhc']['evaluations'])
rhc['time'] = time.time() - start
rhc.set_index('evaluations', inplace=True)
start = time.time()
sa = simulated_annealing(domain=domain,
costf=opt_problem.compute_fitness,
T=experiment_settings['sa']['T'])
sa['time'] = time.time() - start
sa.set_index('temperature', inplace=True)
start = time.time()
ga = genetic_optimize(domain=domain,
costf=opt_problem.compute_fitness,
maxiter=experiment_settings['ga']['generations'])
ga['time'] = time.time() - start
ga.set_index('generations', inplace=True)
start = time.time()
mm = mimic.run_mimic(domain=domain,
fitness_function=opt_problem.compute_fitness,
evaluations=experiment_settings['mm']['evaluations'])
mm['time'] = time.time() - start
return rhc, sa, ga, mm
def compare_all(data, experiment_settings=DEFAULT_EXPERIMENT_SETTINGS):
opt_problem = ClassifierOptimization(data)
domain = opt_problem.domain
start = time.time()
rhc = hillclimb(domain=domain,
costf=opt_problem.compute_classification_error,
max_evaluations=experiment_settings['rhc']['evaluations'])
rhc['optimal_value'] += 1
rhc['time'] = time.time() - start
rhc.set_index('evaluations', inplace=True)
start = time.time()
sa = simulated_annealing(domain=domain,
costf=opt_problem.compute_classification_error,
T=experiment_settings['sa']['T'])
sa.set_index('temperature', inplace=True)
sa['optimal_value'] += 1
sa['time'] = time.time() - start
start = time.time()
ga = genetic_optimize(domain=domain,
costf=opt_problem.compute_classification_error,
maxiter=experiment_settings['ga']['generations'])
ga.set_index('generations', inplace=True)
ga['optimal_value'] += 1
ga['time'] = time.time() - start
start = time.time()
mm = mimic.run_mimic(domain=domain,
fitness_function=opt_problem.compute_classification_error,
evaluations=experiment_settings['mm']['evaluations'])
mm['optimal_value'] += 1
mm['time'] = time.time() - start
return rhc, sa, ga, mm
def evaluate_ga():
domain = [(0, 8)] * len(people) * 2
df = genetic_optimize(domain, schedulecost)
df['optimal_value'] = 1 / df['cost']
return df
def evaluate_ga():
domain = [(0,8)] * len(people) *2
df = genetic_optimize(domain, schedulecost)
df['optimal_value'] = 1 / df['cost']
return df