def test_partial_dependence_display_wrong_len_kind(
pyplot,
clf_diabetes,
diabetes,
):
"""Check that we raise an error when `kind` is a list with a wrong length.
This case can only be triggered using the `PartialDependenceDisplay.from_estimator`
method.
"""
disp = PartialDependenceDisplay.from_estimator(
clf_diabetes,
diabetes.data,
features=[0, 2],
grid_resolution=20,
kind="average", # len(kind) != len(features)
)
# alter `kind` to be a list with a length different from length of `features`
disp.kind = ["average"]
err_msg = (
r"When `kind` is provided as a list of strings, it should contain as many"
r" elements as `features`. `kind` contains 1 element\(s\) and `features`"
r" contains 2 element\(s\)."
)
with pytest.raises(ValueError, match=err_msg):
disp.plot()
from sklearn.inspection import partial_dependence
from sklearn.inspection import PartialDependenceDisplay
print("Computing partial dependence plots...")
tic = time()
features = ["MedInc", "AveOccup", "HouseAge", "AveRooms"]
display = PartialDependenceDisplay.from_estimator(
est,
X_train,
features,
kind="both",
subsample=50,
n_jobs=3,
grid_resolution=20,
random_state=0,
ice_lines_kw={
"color": "tab:blue",
"alpha": 0.2,
"linewidth": 0.5
},
pd_line_kw={
"color": "tab:orange",
"linestyle": "--"
},
)
print(f"done in {time() - tic:.3f}s")
display.figure_.suptitle(
"Partial dependence of house value on non-location features\n"
"for the California housing dataset, with MLPRegressor")
display.figure_.subplots_adjust(hspace=0.3)
from sklearn.inspection import PartialDependenceDisplay
common_params = {
"subsample": 50,
"n_jobs": 2,
"grid_resolution": 20,
"centered": True,
"random_state": 0,
}
print("Computing partial dependence plots...")
tic = time()
display = PartialDependenceDisplay.from_estimator(
est,
X_train,
features=["MedInc", "AveOccup", "HouseAge", "AveRooms"],
kind="both",
**common_params,
)
print(f"done in {time() - tic:.3f}s")
display.figure_.suptitle(
"Partial dependence of house value on non-location features\n"
"for the California housing dataset, with MLPRegressor")
display.figure_.subplots_adjust(hspace=0.3)
# %%
# Gradient boosting
# .................
#
# Let's now fit a :class:`~sklearn.ensemble.HistGradientBoostingRegressor` and
# compute the partial dependence on the same features.
tree.fit(X, y)
mlp.fit(X, y)
# %%
# Plotting partial dependence for two features
# ============================================
#
# We plot partial dependence curves for features "age" and "bmi" (body mass
# index) for the decision tree. With two features,
# :func:`~sklearn.inspection.PartialDependenceDisplay.from_estimator` expects to plot
# two curves. Here the plot function place a grid of two plots using the space
# defined by `ax` .
fig, ax = plt.subplots(figsize=(12, 6))
ax.set_title("Decision Tree")
tree_disp = PartialDependenceDisplay.from_estimator(tree,
X, ["age", "bmi"],
ax=ax)
# %%
# The partial dependence curves can be plotted for the multi-layer perceptron.
# In this case, `line_kw` is passed to
# :func:`~sklearn.inspection.PartialDependenceDisplay.from_estimator` to change the
# color of the curve.
fig, ax = plt.subplots(figsize=(12, 6))
ax.set_title("Multi-layer Perceptron")
mlp_disp = PartialDependenceDisplay.from_estimator(mlp,
X, ["age", "bmi"],
ax=ax,
line_kw={"color": "red"})
# %%
X = np.c_[f_0, f_1]
noise = rng.normal(loc=0.0, scale=0.01, size=n_samples)
y = 5 * f_0 + np.sin(10 * np.pi * f_0) - 5 * f_1 - np.cos(
10 * np.pi * f_1) + noise
fig, ax = plt.subplots()
# Without any constraint
gbdt = HistGradientBoostingRegressor()
gbdt.fit(X, y)
disp = PartialDependenceDisplay.from_estimator(
gbdt,
X,
features=[0, 1],
line_kw={
"linewidth": 4,
"label": "unconstrained",
"color": "tab:blue"
},
ax=ax,
)
# With positive and negative constraints
gbdt = HistGradientBoostingRegressor(monotonic_cst=[1, -1])
gbdt.fit(X, y)
PartialDependenceDisplay.from_estimator(
gbdt,
X,
features=[0, 1],
feature_names=(