def test_generation(self):
# まず初期化する
test_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=self._cross,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function,
)
population_dept = self._db_facade.select_department('populations')
test_ga.generation(steps=20,
experiment_id=1000,
log_span=10,
population_dept=population_dept)
def __init__(self,
situation,
candle_type,
pair,
population,
mutation,
cross,
elite_num,
host,
fitness_function_name,
crossover_name,
hyper_params=None):
"""
:param situation: Situation 遺伝子の要素の取りうる範囲などを表す
:param candle_type: string ロウソク足データの種類
:param pair: string コインの種類
:param population: int 個体数
:param mutation: int 突然変異のパーセンテージ
:param cross: int 交叉のパーセンテージ
:param elite_num: int 世代交代時に残すエリート個体数
:param host: string dbの接続先
:param fitness_function_name: string 適応度関数の名称
:param crossover_name: string 交叉手法の名称
:param hyper_params: dict ハイパーパラメータ
"""
self._db_facade = facade.Facade(host=host)
ff_facade = fffacade.Facade(candle_type=candle_type, coin=pair)
c_facade = cfacade.CrossoverFacade()
expt_logs_dept = self._db_facade.select_department('experiment_logs')
# 適応度関数
fitness_function = ff_facade.select_department(
function_name=fitness_function_name,
db_dept=expt_logs_dept,
hyper_params=hyper_params)
# 交叉手法
crossover_method = c_facade.select_department(
crossover_name=crossover_name)
self._ga = ga.GeneticAlgorithm(mutation=mutation,
cross=cross,
situation=situation,
elite_num=elite_num,
population=population,
crossover_method=crossover_method,
fitness_function=fitness_function)
self._hyper_params = hyper_params
self.__pair = pair
def test_determine_next_generation(self):
"""
世代交代で遺伝子の次元が適当どうかチェックする
"""
# まず初期化する
test_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=self._cross,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function,
)
result = test_ga.init_population()
test_ga.set_geno_type(result)
geno_type = test_ga.determine_next_generation()
self.assertEqual(True, np.all(np.equal(result, geno_type)))
def test_init_population(self):
"""
遺伝子初期化時に次元が適当かどうかのチェック
"""
test_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=self._cross,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function,
)
result = test_ga.init_population()
self.assertIsInstance(result, type(np.asarray([1])))
for i in result:
self.assertLessEqual(i[0], 50)
self.assertLessEqual(1, i[0])
self.assertIsInstance(i, type(np.asarray([])))
self.assertEqual(self._population, len(result))
del test_ga
def test_calc_fitness(self):
"""
適応度の次元が適当かどうかのチェック
"""
test_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=self._cross,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function,
)
result = test_ga.init_population()
test_ga.set_geno_type(result)
fitness = test_ga.calc_fitness(should_log=False)
fitness_list = list()
for i in range(self._population):
fitness_list.append(i + 1)
test_fitness = np.asarray(fitness_list)
self.assertEqual(True, np.all(np.equal(test_fitness, fitness)))
del test_ga
def test__init__(self):
"""
1. mutation(突然変異のパーセンテージ)は0 ~ 100
2. cross(交叉のパーセンテージ)は0 ~ 100
3. situation(遺伝子の特徴量を表す)はmodules.feature.Situationクラスのインスタンス
4. population(個体数)がelite_num(エリートの個体数)より大きい
5. crossover_method(交叉の手法を表す)はmodules.crossover.Crossoverクラスのインスタンス
6. fitness_function(適応度関数を表す)はmodules.fitnessfunction.FitnessFunctionクラスのインスタンス
"""
try:
# mutation(突然変異のパーセンテージ)は0 ~ 100
test_ga = ga.GeneticAlgorithm(
mutation=101,
cross=self._cross,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function)
self.assertEquals(1, -1)
except TypeError:
pass
try:
# mutation(突然変異のパーセンテージ)は0 ~ 100
test_ga = ga.GeneticAlgorithm(
mutation=-1,
cross=self._cross,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function)
self.assertEquals(1, -1)
except TypeError:
pass
try:
# cross(交叉のパーセンテージ)は0 ~ 100
test_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=101,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function)
self.assertEquals(1, -1)
except TypeError:
pass
try:
# cross(交叉のパーセンテージ)は0 ~ 100
test_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=-1,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function)
self.assertEquals(1, -1)
except TypeError:
pass
try:
# situation(遺伝子の特徴量を表す)はmodules.feature.Situationクラスのインスタンス
temp_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=self._cross,
situation=[(1, 50), (50, 100), (1, 50)],
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function)
self.assertEquals(1, -1)
except TypeError:
pass
try:
# population(個体数)がelite_num(エリートの個体数)より大きい
temp_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=self._cross,
situation=self._situation,
elite_num=101,
population=100,
crossover_method=self._crossover_method,
fitness_function=self._fitness_function)
self.assertEquals(1, -1)
except ArithmeticError:
pass
try:
# crossover_method(交叉の手法を表す)はmodules.crossover.Crossoverクラスのインスタンス
test_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=self._cross,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=100,
fitness_function=self._fitness_function)
self.assertEquals(1, -1)
except TypeError:
pass
try:
# fitness_function(適応度関数を表す)はmodules.fitnessfunction.FitnessFunctionクラスのインスタンス
test_ga = ga.GeneticAlgorithm(
mutation=self._mutation,
cross=self._cross,
situation=self._situation,
elite_num=self._elite_num,
population=self._population,
crossover_method=self._crossover_method,
fitness_function=100,
)
self.assertEquals(1, -1)
except TypeError:
pass