def test_feature_onehot_not_exist(self, titanic_data):
with pytest.raises(ValueError):
_ = _check_feature(
feature=[
"Embarked_C", "Embarked_S", "Embarked_Q", "Embarked_F"
],
df=titanic_data,
)
def plot_pdp(
X: pd.DataFrame,
y: np.array,
model,
var_name: str,
target_name: str,
num_grid_points: int = 10,
) -> go.Figure:
## Concat target column
df = pd.concat([X, pd.DataFrame({target_name: y})], axis=1)
## Compute mean target per bin
data_x, _, summary_df, _, _, _ = _prepare_info_plot_data(
feature=var_name,
feature_type=_check_feature(var_name, df),
data=df,
num_grid_points=num_grid_points,
grid_type="percentile",
percentile_range=None,
grid_range=None,
cust_grid_points=None,
show_percentile=False,
show_outliers=False,
endpoint=True,
)
target_line = (data_x.groupby("x", as_index=False).agg({
target_name: "mean"
}).sort_values("x", ascending=True))
summary_df = summary_df.merge(target_line, on="x", how="outer")[[
"display_column", target_name
]].rename({
"display_column": var_name,
target_name: "mean_target"
}, axis=1)
## Compute Partial dependence plot
pdp_array = pdp.pdp_isolate(
model=model,
dataset=X,
model_features=X.columns,
feature=var_name,
num_grid_points=num_grid_points,
).pdp
pdp_array = 0.5 * (pdp_array[1:] + pdp_array[:-1])
## Merge with summary_df
summary_df["pdp"] = pdp_array
## Plot figure
fig = go.Figure()
fig.add_trace(
go.Scatter(
x=summary_df[var_name],
y=summary_df["mean_target"],
mode="lines",
name="mean_target",
line=dict(color="rgb(46, 186, 175)", width=4),
))
fig.add_trace(
go.Scatter(
x=summary_df[var_name],
y=summary_df["pdp"],
mode="lines",
name="partial_dependence_plot",
line=dict(color="rgb(255, 161, 105)", width=4),
))
title = "Mean conversion rate and Partial dependence plot"
fig.update_layout(
title={
"text": title,
"y": 0.9,
"x": 0.5,
"xanchor": "center",
"yanchor": "top",
},
legend=dict(x=0.65, y=0.9),
xaxis_title=var_name,
yaxis_title=target_name,
yaxis_range=[0, 0.25],
autosize=False,
width=800,
height=500,
plot_bgcolor="rgb(240, 240, 240)",
)
return fig.show()
def test_feature_onehot_incomplete(self, titanic_data):
with pytest.raises(ValueError):
_ = _check_feature(feature=["Embarked_C"], df=titanic_data)
def test_feature_not_exist(self, titanic_data):
with pytest.raises(ValueError):
_ = _check_feature(feature="gender", df=titanic_data)
def test_feature_onehot(self, titanic_data):
feature_type = _check_feature(
feature=["Embarked_C", "Embarked_S", "Embarked_Q"],
df=titanic_data)
assert feature_type == "onehot"
def test_feature_numeric(self, titanic_data):
feature_type = _check_feature(feature="Fare", df=titanic_data)
assert feature_type == "numeric"
def test_feature_binary(self, titanic_data):
feature_type = _check_feature(feature="Sex", df=titanic_data)
assert feature_type == "binary"
def test_feature_onehot_not_exist(self, titanic_data):
with pytest.raises(ValueError):
_ = _check_feature(feature=['Embarked_C', 'Embarked_S', 'Embarked_Q', 'Embarked_F'], df=titanic_data)
def test_feature_numeric(self, titanic_data):
feature_type = _check_feature(feature='Fare', df=titanic_data)
assert feature_type == 'numeric'
def test_feature_onehot(self, titanic_data):
feature_type = _check_feature(feature=['Embarked_C', 'Embarked_S', 'Embarked_Q'], df=titanic_data)
assert feature_type == 'onehot'
def test_feature_binary(self, titanic_data):
feature_type = _check_feature(feature='Sex', df=titanic_data)
assert feature_type == 'binary'
