def plot_components(self, X_true=None, y_true=None, groups=None, fig=None):
import matplotlib.pyplot as plt
if fig is None:
fig = plt.figure(figsize=(18, 1))
lookahead_scale = 0.3
t_min, t_max = self._X_scaler_.min_["t"], self._X_scaler_.max_["t"]
t_max += (t_max - t_min) * lookahead_scale
t = pd.date_range(t_min, t_max, freq="D")
scaled_t = np.linspace(0, 1 + lookahead_scale, len(t))
total = self.plot(self.trace_, scaled_t, self._y_scaler_)
ax = add_subplot()
ax.set_title("overall")
ax.plot(t, self._y_scaler_.inv_transform(total))
if X_true is not None and y_true is not None:
if groups is not None:
for group in groups.cat.categories:
mask = groups == group
ax.scatter(
X_true["t"][mask],
y_true[mask],
label=group,
marker=".",
alpha=0.2,
)
else:
ax.scatter(X_true["t"], y_true, c="k", marker=".", alpha=0.2)
fig.tight_layout()
return fig
def plot(self, trace, scaled_t, y_scaler):
scaled_s = pd.Series(self._predict(trace, scaled_t).mean(axis=1), name='value')
s = y_scaler.inv_transform(scaled_s)
ax = add_subplot()
ax.set_title(str(self))
ax.set_xticks([])
ax.plot(scaled_t, s, c="lightblue")
return scaled_s
def plot(self, trace, scaled_t, y_scaler):
ax = add_subplot()
ax.set_title(str(self))
trend_return = np.empty((len(scaled_t), len(self.groups_)))
plot_data = []
for group_code, group_name in self.groups_.items():
y_hat = np.mean(self._predict(trace, scaled_t, group_code), axis=1)
trend_return[:, group_code] = y_hat
plot_data.append((group_name, y_hat[0]))
ax.bar(*zip(*plot_data))
ax.axhline(0, c="k", linewidth=3)
return trend_return
def plot(self, trace, scaled_t, y_scaler):
ax = add_subplot()
scaled_trend = self._predict(trace, scaled_t)
trend = y_scaler.inv_transform(scaled_trend)
ax.set_title(str(self))
ax.set_xticks([])
ax.plot(scaled_t, trend.mean(axis=1), c="lightblue")
for changepoint in self.s:
ax.axvline(changepoint, linestyle="--", alpha=0.2, c="k")
return scaled_trend.mean(axis=1)
def plot(self, trace, scaled_t, y_scaler):
ax = add_subplot()
ax.set_title(str(self))
ax.set_xticks([])
trend_return = np.empty((len(scaled_t), len(self.groups_)))
for group_code, group_name in self.groups_.items():
y_hat = mode(self._predict(trace, scaled_t, group_code),
axis=1)[0][:, 0]
ax.plot(scaled_t, y_hat, label=group_name)
trend_return[:, group_code] = y_hat
ax.set_ylim([-1.05, 1.05])
ax.legend()
return trend_return
def plot(self, trace, scaled_t, y_scaler):
ax = add_subplot()
ax.set_title(str(self))
seasonality_return = np.empty((len(scaled_t), len(self.groups_)))
for group_code, group_name in self.groups_.items():
scaled_s = self._predict(trace, scaled_t, group_code)
s = y_scaler.inv_transform(scaled_s)
ax.plot(list(range(self.period.days)), s.mean(axis=1)[:self.period.days], label=group_name)
seasonality_return[:, group_code] = scaled_s.mean(axis=1)
return seasonality_return
def plot(self, trace, scaled_t, y_scaler):
ax = add_subplot()
ax.set_title(str(self))
ax.set_xticks([])
trend_return = np.empty((len(scaled_t), len(self.groups_)))
for group_code, group_name in self.groups_.items():
scaled_trend = self._predict(trace, scaled_t, group_code)
trend = y_scaler.inv_transform(scaled_trend)
ax.plot(scaled_t, trend.mean(axis=1), label=group_name)
trend_return[:, group_code] = scaled_trend.mean(axis=1)
for changepoint in self.s:
ax.axvline(changepoint, linestyle="--", alpha=0.2, c="k")
ax.legend()
return trend_return