def test_model_persistence(mocker):
# Simple single task SISSO runs with various options for the run directory
# Mock the run of the custodian by just copying a reference SISSO.out file
def copy_sisso_out():
shutil.copy(
os.path.join(TEST_FILES_DIR, "runs", "perfect_linear_5pts",
"SISSO.out"),
"SISSO.out",
)
mocker.patch.object(
pysisso.sklearn.Custodian,
"run",
return_value=[],
side_effect=copy_sisso_out,
)
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1, rung=0, subs_sis=1, method="L0")
sisso_reg.fit(np.array([[1], [2], [3], [4], [5]]),
np.array([0, 1, 2, 3, 4]))
joblib.dump(sisso_reg, filename="model.joblib")
sisso_reg_loaded = joblib.load("model.joblib")
pred = sisso_reg_loaded.predict([[1.5], [4.5]])
assert pred[0] == 0.5
assert pred[1] == 3.5
assert sisso_reg.get_params() == sisso_reg_loaded.get_params()
model = sisso_reg.sisso_out.model
model_loaded = sisso_reg_loaded.sisso_out.model
assert len(model.coefficients) == 1
assert len(model_loaded.coefficients) == 1
assert model.coefficients[0] == pytest.approx(
model_loaded.coefficients[0])
def test_sisso_regressor_omp(mocker):
# Simple SISSO run with OMP
# Mock the run of the custodian by just copying a reference SISSO.out file
def copy_sisso_out():
shutil.copy(
os.path.join(TEST_FILES_DIR, "runs", "OMP", "SISSO.out"),
"SISSO.out",
)
mocker.patch.object(
pysisso.sklearn.Custodian,
"run",
return_value=[],
side_effect=copy_sisso_out,
)
with ScratchDir("."):
sisso_reg = SISSORegressor.OMP(desc_dim=4)
assert sisso_reg.rung == 0
assert sisso_reg.subs_sis == 1
assert sisso_reg.desc_dim == 4
assert sisso_reg.method == "L0"
assert sisso_reg.L1L0_size4L0 is None
X = np.array([
[8, 1, 3.01, 4],
[6, 2, 3.02, 3],
[2, 3, 3.01, 0],
[10, 4, 3.02, -8],
[4, 5, 3.01, 10],
])
y = 0.9 * X[:, 1] + 0.1 * X[:, 3] - 1.0
sisso_reg.fit(X, y)
actual_sin = "SISSO_dir/SISSO.in"
ref_sin = os.path.join(TEST_FILES_DIR, "runs", "OMP", "SISSO.in")
assert [line for line in open(actual_sin)
] == [line for line in open(ref_sin)]
sisso_out = SISSOOut.from_file(filepath="SISSO_dir/SISSO.out")
assert sisso_out.params.n_rungs == sisso_reg.rung
assert sisso_out.params.SIS_subspaces_sizes == [sisso_reg.subs_sis]
assert sisso_out.params.descriptor_dimension == sisso_reg.desc_dim
assert sisso_out.params.sparsification_method == sisso_reg.method
sisso_model = sisso_out.model
assert str(sisso_model.descriptors[0]) == "(feature_1)"
assert str(sisso_model.descriptors[1]) == "(feature_3)"
from pysisso.outputs import SISSOOut
from pysisso.sklearn import SISSORegressor
# Define the data set
X = np.array([
[8, 1, 3.01, 4],
[6, 2, 3.02, 3],
[2, 3, 3.01, 0],
[10, 4, 3.02, -8],
[4, 5, 3.01, 10],
])
y = 0.9 * X[:, 1] + 0.1 * X[:, 3] - 1.0
# Define the regressor and fit the data
sisso_reg = SISSORegressor.OMP(desc_dim=4)
sisso_reg.fit(X,
y,
columns=["feature_0", "feature_1", "feature_2", "feature_3"])
# Get the final model obtained
sisso_out = SISSOOut.from_file(filepath="SISSO_dir/SISSO.out")
sisso_model = sisso_out.model
# Get the descriptors
descriptors = [str(d) for d in sisso_model.descriptors]
# Print the order of the OMP features
# Should start with feature_1, then feature_3.
# feature_0 and feature_2 might be interchanged.
for idesc, desc in enumerate(descriptors):
fig, subplot = plt.subplots()
subplot.plot(xlin, ylin, "-", color="C0", label="True function")
subplot.plot(X, y, "o", color="C1", label="Data")
subplot.set_xlabel("x")
subplot.set_ylabel("f(x)")
subplot.set_title(TITLE)
subplot.legend()
if SAVE_FIGURES:
fig.savefig("true_data.pdf")
if PLOT_FIGURES:
plt.show()
# Define the regressor, fit the data and predict
sisso_regressor = SISSORegressor(
rung=1,
opset="(+)(*)(^2)(^3)(^-1)(cos)(sin)",
desc_dim=3,
clean_run_dir=CLEAN_RUN_DIR,
)
X = X.reshape(
-1, 1) # only one feature, X is initially defined as 1D, sklearn needs 2D
sisso_regressor.fit(X, y)
ylin_pred = sisso_regressor.predict(xlin)
# Plot the true and predicted functions, together with the data
fig, subplot = plt.subplots()
subplot.plot(xlin, ylin, "-", color="C0", label="True function")
subplot.plot(X, y, "o", color="C1", label="Data")
subplot.plot(xlin, ylin_pred, "-", color="C2", label="Predicted function")
subplot.set_xlabel("x")
subplot.set_ylabel("f(x)")
subplot.set_title(TITLE)
def test_sisso_regressor(mocker):
# Simple single task SISSO runs with various options for the run directory
# Mock the run of the custodian by just copying a reference SISSO.out file
def copy_sisso_out():
shutil.copy(
os.path.join(TEST_FILES_DIR, "runs", "perfect_linear_5pts",
"SISSO.out"),
"SISSO.out",
)
mocker.patch.object(
pysisso.sklearn.Custodian,
"run",
return_value=[],
side_effect=copy_sisso_out,
)
makedirs_spy = mocker.spy(pysisso.sklearn, "makedirs_p")
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1, rung=0, subs_sis=1, method="L0")
sisso_reg.fit(np.array([[1], [2], [3], [4], [5]]),
np.array([0, 1, 2, 3, 4]))
pred = sisso_reg.predict([[1.5], [4.5]])
assert pred[0] == 0.5
assert pred[1] == 3.5
assert os.path.exists("SISSO_dir")
makedirs_spy.assert_called_with("SISSO_dir")
assert makedirs_spy.call_count == 1
makedirs_spy.reset_mock()
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1,
rung=0,
subs_sis=1,
method="L0",
run_dir="mySISSOdir")
sisso_reg.fit(np.array([[1], [2], [3], [4], [5]]),
np.array([0, 1, 2, 3, 4]))
pred = sisso_reg.predict([[1.5], [4.5]])
assert pred[0] == 0.5
assert pred[1] == 3.5
assert os.path.exists("mySISSOdir")
assert not os.path.exists("SISSO_dir")
makedirs_spy.assert_called_with("mySISSOdir")
assert makedirs_spy.call_count == 1
makedirs_spy.reset_mock()
with ScratchDir("."):
sisso_reg = SISSORegressor(
desc_dim=1,
rung=0,
subs_sis=1,
method="L0",
run_dir="mySISSOdir",
clean_run_dir=True,
)
sisso_reg.fit(np.array([[1], [2], [3], [4], [5]]),
np.array([0, 1, 2, 3, 4]))
pred = sisso_reg.predict([[1.5], [4.5]])
assert pred[0] == 0.5
assert pred[1] == 3.5
assert not os.path.exists("mySISSOdir")
makedirs_spy.assert_called_with("mySISSOdir")
assert makedirs_spy.call_count == 1
makedirs_spy.reset_mock()
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1,
rung=0,
subs_sis=1,
method="L0",
clean_run_dir=True)
sisso_reg.fit(np.array([[1], [2], [3], [4], [5]]),
np.array([0, 1, 2, 3, 4]))
pred = sisso_reg.predict([[1.5], [4.5]])
assert pred[0] == 0.5
assert pred[1] == 3.5
assert not os.path.exists("SISSO_dir")
makedirs_spy.assert_called_with("SISSO_dir")
assert makedirs_spy.call_count == 1
makedirs_spy.reset_mock()
# Run with a temporary directory (i.e. when run_dir is None, useful for CV)
# TODO : mocking tempfile did not work here for some reason ...
mocker.patch(
"pysisso.sklearn.get_timestamp",
return_value="2018_09_28_16_04_54_017895",
)
with ScratchDir("."):
sisso_reg = SISSORegressor(
desc_dim=1,
rung=0,
subs_sis=1,
method="L0",
run_dir=None,
clean_run_dir=False,
)
sisso_reg.fit(np.array([[1], [2], [3], [4], [5]]),
np.array([0, 1, 2, 3, 4]))
pred = sisso_reg.predict([[1.5], [4.5]])
assert pred[0] == 0.5
assert pred[1] == 3.5
assert os.path.exists("SISSO_runs")
dirs = os.listdir("SISSO_runs")
assert len(dirs) == 1
sisso_dir = dirs[0]
assert sisso_dir.startswith("SISSO_dir_2018_09_28_16_04_54_017895_")
makedirs_spy.assert_called_with("SISSO_runs")
assert makedirs_spy.call_count == 1
makedirs_spy.reset_mock()
# Run with a temporary directory (i.e. when run_dir is None, useful for CV)
with ScratchDir("."):
sisso_reg = SISSORegressor(
desc_dim=1,
rung=0,
subs_sis=1,
method="L0",
run_dir=None,
clean_run_dir=True,
)
sisso_reg.fit(np.array([[1], [2], [3], [4], [5]]),
np.array([0, 1, 2, 3, 4]))
pred = sisso_reg.predict([[1.5], [4.5]])
assert pred[0] == 0.5
assert pred[1] == 3.5
assert os.path.exists("SISSO_runs")
dirs = os.listdir("SISSO_runs")
assert len(dirs) == 0
assert makedirs_spy.call_count == 1
makedirs_spy.reset_mock()
# Simple multi task SISSO run
# Mock the run of the custodian by just copying a reference SISSO.out file
def copy_sisso_out():
shutil.copy(
os.path.join(TEST_FILES_DIR, "runs", "perfect_linear_5pts_multi",
"SISSO.out"),
"SISSO.out",
)
mocker.patch.object(
pysisso.sklearn.Custodian,
"run",
return_value=[],
side_effect=copy_sisso_out,
)
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1, rung=0, subs_sis=1, method="L0")
sisso_reg.fit(
np.array([[1], [2], [3], [4], [5]]),
np.array([[0, -3], [1, -5], [2, -7], [3, -9], [4, -11]]),
)
pred = sisso_reg.predict([[1.5], [4.5]])
assert pred[0] == pytest.approx([0.5, -4])
assert pred[1] == pytest.approx([3.5, -10])
assert sisso_reg.columns == ["feat1"]
# Test of initializations and errors
# Run with a numpy array
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1, rung=0, subs_sis=1, method="L0")
sisso_reg.fit(
np.array([[1, 5], [2, 3], [3, 89], [4, 1], [5, 4]]),
np.array([[0, -3], [1, -5], [2, -7], [3, -9], [4, -11]]),
)
assert sisso_reg.columns == ["feat1", "feat2"]
# Run with a pandas Dataframe
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1, rung=0, subs_sis=1, method="L0")
X_df = pd.DataFrame([[1, 5], [2, 3], [3, 89], [4, 1], [5, 4]],
columns=["a", "b"])
sisso_reg.fit(X_df,
np.array([[0, -3], [1, -5], [2, -7], [3, -9], [4, -11]]))
assert sisso_reg.columns == ["a", "b"]
# Run raising errors about columns
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1, rung=0, subs_sis=1, method="L0")
X_df = pd.DataFrame([[1, 5], [2, 3], [3, 89], [4, 1], [5, 4]],
columns=["a", "b"])
with pytest.raises(
ValueError,
match=r"Columns should be of the size of the "
r"second axis of X.",
):
sisso_reg.fit(
X_df,
np.array([[0, -3], [1, -5], [2, -7], [3, -9], [4, -11]]),
columns=["a", "b", "c"],
)
# Run raising errors about index
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1, rung=0, subs_sis=1, method="L0")
X_df = pd.DataFrame([[1], [2], [3], [4], [5]])
with pytest.raises(ValueError,
match=r"Index, X and y should have same size."):
sisso_reg.fit(X_df, np.array([[0], [1], [2], [3]]))
# Run raising errors about index
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1, rung=0, subs_sis=1, method="L0")
X_df = pd.DataFrame([[1], [2], [3], [4], [5]])
with pytest.raises(ValueError,
match=r"Index, X and y should have same size."):
sisso_reg.fit(
X_df,
np.array([[0], [1], [2], [3], [4]]),
index=["a", "b", "c", "d", "e", "f"],
)
# Run with a wrong shape for y target
with ScratchDir("."):
sisso_reg = SISSORegressor(desc_dim=1, rung=0, subs_sis=1, method="L0")
X_df = pd.DataFrame([[1], [2], [3], [4], [5]])
with pytest.raises(ValueError, match=r"Wrong shapes."):
sisso_reg.fit(X_df, np.array([[[0], [1], [2], [3], [4]]]))
subplot.set_ylabel("f(x)")
subplot.set_title(TITLE)
subplot.legend()
if SAVE_FIGURES:
fig.savefig("true_data.pdf")
if PLOT_FIGURES:
plt.show()
# Define the regressor and the grid search, fit the data and predict
# Note that run_dir HAS to be None here, so that concurrent SISSO runs from different
# folds and/or hyperparameters sets do not interfere with one another
sisso_regressor = SISSORegressor(
rung=1,
opset="(+)(-)(*)(^2)(^3)(^-1)(exp)(sin)(cos)",
desc_dim=3,
subs_sis=40,
method="L1L0",
L1L0_size4L0=15,
run_dir=None,
clean_run_dir=CLEAN_RUN_DIR,
)
param_grid = {"rung": [0, 1, 2], "desc_dim": [2, 3, 4]}
grid_search = GridSearchCV(
estimator=sisso_regressor, param_grid=param_grid, cv=4, n_jobs=4
)
X = X.reshape(-1, 1) # only one feature, X is initially defined as 1D, sklearn needs 2D
grid_search.fit(X, y)
ylin_pred = grid_search.predict(xlin)
# Plot the true and predicted functions, together with the data
fig, subplot = plt.subplots()
subplot.plot(xlin, ylin, "-", color="C0", label="True function")