def test_prediction_mean(monkeypatch):
params = {
'data': {
'ew_lags': 12.166882847847372,
'returns_lags': 5
},
'model': {
'bagging_temperature': 1.5889359773128047,
'depth': 2,
'ignored_features': (),
'l2_leaf_reg': 3.0691833972589424,
'learning_rate': 0.04230361495762283,
'one_hot_max_size': 2,
'random_strength': 3.6085599262888683
}
}
monkeypatch.setattr(model.ReturnsModel, 'PARAMS', params)
returns = model.ReturnsModel(('MSRS', 'MSTT', 'RTKMP', 'UPRO', 'FEES'),
pd.Timestamp('2018-10-11')).prediction_mean
assert isinstance(returns, pd.Series)
assert len(returns) == 5
assert returns.index.tolist() == ['MSRS', 'MSTT', 'RTKMP', 'UPRO', 'FEES']
assert returns['MSRS'] == 0.0018521857721484048
assert returns['MSTT'] == 0.004415906411386809
assert returns['RTKMP'] == 0.002905142786419567
assert returns['UPRO'] == 0.0028195209196578627
assert returns['FEES'] == 0.0037373919332427726
def test_prediction_std(monkeypatch):
params = {
'data': {
'ew_lags': 12.166882847847372,
'returns_lags': 5
},
'model': {
'bagging_temperature': 1.5889359773128047,
'depth': 2,
'ignored_features': (),
'l2_leaf_reg': 3.0691833972589424,
'learning_rate': 0.04230361495762283,
'one_hot_max_size': 2,
'random_strength': 3.6085599262888683
}
}
monkeypatch.setattr(model.ReturnsModel, 'PARAMS', params)
std = model.ReturnsModel(('MSRS', 'RTKM', 'RTKMP', 'UPRO', 'FEES'),
pd.Timestamp('2018-10-11')).prediction_std
assert isinstance(std, pd.Series)
assert len(std) == 5
assert std.index.tolist() == ['MSRS', 'RTKM', 'RTKMP', 'UPRO', 'FEES']
assert std['MSRS'] == 0.049892858163254374
assert std['RTKM'] == 0.04798205289008257
assert std['RTKMP'] == 0.03734821638483238
assert std['UPRO'] == 0.046223684807991286
assert std['FEES'] == 0.08119363794137813
def test_new_manager():
positions = ('LSRG', 'VSMO', 'AKRN', 'GCHE', 'MTSS')
date = pd.Timestamp('2018-10-12')
data = manager.ReturnsMLDataManager(positions, date)
assert data.value.params == model.ReturnsModel(positions, date).params
assert data.value.positions == positions
assert data.value.date == date
def test_make_data_space():
returns = model.ReturnsModel(('CHMF', 'MSTT', 'RTKMP', 'UPRO', 'FEES'),
pd.Timestamp('2018-10-11'))
space = returns._make_data_space()
assert isinstance(space, dict)
assert len(space) == 2
assert isinstance(space['ew_lags'], hyperopt.pyll.Apply)
assert isinstance(space['returns_lags'], hyperopt.pyll.Apply)
def test_check_data_space_bounds_lower(capsys):
returns = model.ReturnsModel(('CHMF', 'MSTT', 'RTKMP', 'UPRO', 'FEES'),
pd.Timestamp('2018-10-11'))
params = copy.deepcopy(model.PARAMS)
params['data']['ew_lags'] /= 1 + model.EW_LAGS_RANGE * 0.91
returns._check_data_space_bounds(params)
captured = capsys.readouterr()
assert 'Необходимо увеличить EW_LAGS_RANGE до' in captured.out
assert 'Необходимо увеличить MAX_RETURNS_LAGS до' not in captured.out
def test_manager_change_params(monkeypatch):
fake_params = copy.deepcopy(model.ReturnsModel.PARAMS)
fake_params['data']['returns_lags'] += 1
monkeypatch.setattr(model.ReturnsModel, 'PARAMS', fake_params)
positions = ('LSRG', 'VSMO', 'MSRS', 'GCHE', 'MTSS')
date = pd.Timestamp('2018-10-11')
data = manager.ReturnsMLDataManager(positions, date)
assert data.value.params == model.ReturnsModel(positions, date).params
assert data.value.params['data']['returns_lags'] == fake_params['data']['returns_lags']
assert data.value.positions == positions
assert data.value.date == date
def test_check_data_space_bounds(capsys):
returns = model.ReturnsModel(('CHMF', 'MSTT', 'RTKMP', 'UPRO', 'FEES'),
pd.Timestamp('2018-10-11'))
returns._check_data_space_bounds(model.PARAMS)
captured = capsys.readouterr()
assert captured.out == ''