def __init__(self, global_hyperparams, hp_grid=None, n=1):
BaseAlgo.__init__(self, global_hyperparams, hp_grid)
self.name = 'Momentum'
self.algo_type = 'TA'
self.n = n
if global_hyperparams['output_type'] == 'R':
raise AlgoError('You cannot call MOM as a regressor')
def __init__(self, global_hyperparams, hp_grid=None, **hyperparams):
BaseAlgo.__init__(self, global_hyperparams, hp_grid)
if global_hyperparams['output_type'] == 'C':
self.model = RandomForestClassifier()
self.name = 'Random Forest Classifier'
else:
self.model = RandomForestRegressor()
self.name = 'Random Forest Regressor'
self.algo_type = 'ML'
self.set_hyperparams(**hyperparams)
def __init__(self, env, iterations, lamb, monte_carlo, save_mse_vals=False):
BaseAlgo.__init__(self, env, iterations)
self.gamma = 0.8
self.lamb = lamb
self.mc = monte_carlo
if save_mse_vals:
self.save_mse_vals = True
self.mse_vals = []
else:
self.save_mse_vals = False
def __init__(self, global_hyperparams, hp_grid=None, **hyperparams):
BaseAlgo.__init__(self, global_hyperparams, hp_grid)
if global_hyperparams['output_type'] == 'C':
self.model = MLPClassifier()
self.name = 'Multi Layer Perceptron Classifier'
else:
self.model = MLPRegressor()
self.name = 'Multi Layer Perception Regressor'
self.algo_type = 'ML'
self.set_hyperparams(**hyperparams)
def __init__(self, global_hyperparams, hp_grid=None, ob=80, os=20, w=None):
BaseAlgo.__init__(self, global_hyperparams, hp_grid)
self.name = 'Relative Strength Index'
self.algo_type = 'TA'
self.ob = ob # The overbought level for the RSI
self.os = os # The oversold level for the RSI
self.w = w if w is not None else global_hyperparams[
'rolling_window_size'] # The window size of the RS
if global_hyperparams['output_type'] == 'R':
raise AlgoError('You cannot call RSI as a regressor')
def __init__(self,
global_hyperparams,
hp_grid=None,
mean_type='arithmetic',
window_size=None):
BaseAlgo.__init__(
self, global_hyperparams, hp_grid
) # allow to run the init of the BaseAlgo class, and define all default arguments
self.name = 'Historical Mean'
self.algo_type = 'ML' # By convention
self.mean_type = mean_type
if window_size is not None: # It is possible to define a window size different from the global rolling window size, but it has to be less or equal
self.window_size = window_size
else:
self.window_size = self.global_hyperparams['rolling_window_size']
def __init__(self,
global_hyperparams,
hp_grid=None,
stw=50,
ltw=200,
a=None,
b=None,
c=None):
BaseAlgo.__init__(self, global_hyperparams, hp_grid)
self.algo_type = 'TA'
self.name = 'Golden Dead Cross'
self.a = a # a>=0, it controls the min ratio of mz/z, a=0: always accept
self.b = b # b>=0, it controls the impact of the criteria on mz, b=0: always accept
self.c = c # c>=0, it controls the max values of z and the min value of mz
self.stw = stw
self.ltw = ltw
self.mz = None # absolute value of the last peak
if global_hyperparams['output_type'] == 'R':
raise AlgoError('You cannot call GDC as a regressor')
class TestEasy21(unittest.TestCase):
def setUp(self):
self.algo1 = BaseAlgo(None, 5, (2, 3))
self.algo2 = BaseAlgo(None, (5, 2), (3, ))
self.algo3 = BaseAlgo(None, (5, 2), (3, 2, 3))
def test_get_action(self):
state = tuple([0] * len(self.algo1.state_space_shape))
self.assertEqual(len(self.algo1.get_action(state)),
len(self.algo1.action_space_shape))
state = tuple([0] * len(self.algo2.state_space_shape))
self.assertEqual(len(self.algo2.get_action(state)),
len(self.algo2.action_space_shape))
state = tuple([0] * len(self.algo3.state_space_shape))
self.assertEqual(len(self.algo3.get_action(state)),
len(self.algo3.action_space_shape))
def test_get_V(self):
pass
def __init__(self,
global_hyperparams,
hp_grid=None,
regularization=None,
**hyperparams):
BaseAlgo.__init__(
self, global_hyperparams, hp_grid, **hyperparams
) # allow to run the init of the BaseAlgo class, and define all default arguments
if regularization is None:
self.model = LinearRegression()
self.name = 'Linear Regression'
elif regularization == 'Lasso':
self.model = Lasso(normalize=True)
self.name = 'Lasso'
elif regularization == 'Ridge':
self.model = Ridge(normalize=True)
self.name = 'Ridge'
elif regularization == 'ElasticNet':
self.model = ElasticNet(normalize=True)
self.name = 'Elastic Net'
self.algo_type = 'ML'
self.set_hyperparams(
**hyperparams
) # Do not forget to fix hyperparams after defining the model
def __init__(self, global_hyperparams, weights):
BaseAlgo.__init__(self, global_hyperparams)
self.algo_type = 'BA'
self.weights = weights
self.name = 'Core Manual Weighting'
def __init__(self, global_hyperparams, scoring):
BaseAlgo.__init__(self, global_hyperparams)
self.algo_type = 'BA'
self.scoring = scoring
self.index_best = None
self.name = 'Best In Sample'
def setUp(self):
self.algo1 = BaseAlgo(None, 5, (2, 3))
self.algo2 = BaseAlgo(None, (5, 2), (3, ))
self.algo3 = BaseAlgo(None, (5, 2), (3, 2, 3))