def partial_dependence_curve(tickers: Tuple[str, ...], date: pd.Timestamp):
"""Рисует кривые частичной зависимости для численных параметров.
:param tickers:
Тикеры, для которых необходимо составить ML-модель.
:param date:
Дата, на которую составляется ML-модель.
"""
params = config.ML_PARAMS
cases = examples.Examples(tickers, date, params["data"])
clf, train_pool_params = train_clf(cases, params)
n_plots = len(train_pool_params["data"].columns) - len(
cases.categorical_features())
axs = axs_iter(n_plots)
results = []
for n, name in enumerate(train_pool_params["data"]):
if n in cases.categorical_features():
continue
ax = next(axs)
pool_params = copy.deepcopy(train_pool_params)
quantiles = pool_params["data"].iloc[:, n].quantile(QUANTILE).values
y = []
for quantile in quantiles:
pool_params["data"].iloc[:, n] = quantile
predict_pool = catboost.Pool(**pool_params)
raw_prediction = clf.predict(predict_pool)
prediction = (raw_prediction * pool_params["data"].iloc[:, 0] *
YEAR_IN_TRADING_DAYS)
y.append(prediction.values.mean())
ax.set_title(f"{name}")
ax.tick_params(labelleft=True)
ax.plot(quantiles, y)
results.append((quantiles, y))
plt.show()
return results
def test_cv_model_raise_max_iter(monkeypatch):
fake_tech_params = dict(**cv.TECH_PARAMS)
fake_max_iter = 3
fake_tech_params["iterations"] = fake_max_iter
monkeypatch.setattr(cv, "TECH_PARAMS", fake_tech_params)
monkeypatch.setattr(cv, "MAX_ITERATIONS", fake_max_iter)
data = examples.Examples(("LSNGP", "LKOH", "GMKN"),
pd.Timestamp("2018-12-14"), PARAMS["data"])
with pytest.raises(POptimizerError) as error:
cv.valid_model(PARAMS, data)
assert "Необходимо увеличить MAX_ITERATIONS =" in str(error.value)
def test_valid_model():
data = examples.Examples(("LSNGP", "LKOH", "GMKN"),
pd.Timestamp("2018-12-14"), PARAMS["data"])
result = cv.valid_model(PARAMS, data)
assert isinstance(result, dict)
assert len(result) == 6
assert result["loss"] == pytest.approx(0.014_495_100_438_051_8)
assert result["status"] == "ok"
assert result["std"] == pytest.approx(0.160839952004336)
assert result["r2"] == pytest.approx(-0.014_495_100_438_051_8)
assert result["data"] == PARAMS["data"]
for key, value in PARAMS["model"].items():
assert result["model"][key] == value
for key, value in cv.TECH_PARAMS.items():
if key == "iterations":
assert result["model"][key] < value
else:
assert result["model"][key] == value
def make_cases(monkeypatch):
return examples.Examples(
("SNGSP", "VSMO", "DSKY"), pd.Timestamp("2018-12-14"), PARAMS["data"]
)
def test_optimize_hyper(monkeypatch, capsys):
space = {
"data": (
("Label", {
"days": hp.choice("label", list(range(21, 31)))
}),
("STD", {
"days": 186
}),
("Ticker", {
"on_off": False
}),
("Mom12m", {
"days": 279
}),
("DivYield", {
"days": 252,
"periods": 1
}),
),
"model": {
"bagging_temperature": 1,
"depth": 6,
"l2_leaf_reg": 3,
"learning_rate": 0.1,
"one_hot_max_size": 2,
"random_strength": 1,
"ignored_features": [],
},
}
cases = examples.Examples(("LSNGP", "LKOH", "GMKN"),
pd.Timestamp("2018-12-14"), PARAMS["data"])
monkeypatch.setattr(cv, "MAX_SEARCHES", 10)
monkeypatch.setattr(cases, "get_params_space", lambda: space["data"])
monkeypatch.setattr(cv, "get_model_space", lambda: space["model"])
result = cv.optimize_hyper(cases)
captured = capsys.readouterr()
assert "Необходимо расширить" in captured.out
assert isinstance(result, dict)
assert len(result) == 2
assert result["data"] == (
("Label", {
"days": 29
}),
("STD", {
"days": 186
}),
("Ticker", {
"on_off": False
}),
("Mom12m", {
"days": 279
}),
("DivYield", {
"days": 252,
"periods": 1
}),
)
model_params = {
"bagging_temperature": 1,
"depth": 6,
"l2_leaf_reg": 3,
"learning_rate": 0.1,
"one_hot_max_size": 2,
"random_strength": 1,
}
for k, v in model_params.items():
assert result["model"][k] == pytest.approx(v)
def create_examples():
yield examples.Examples(("AKRN", "CHMF", "BANEP"),
pd.Timestamp("2018-12-13"), FEAT_PARAMS)