canari.data_visualization#

Visualization for Canari’s results.

This module provides functions to plot:

Raw or standardized data
Prediction results with uncertainty (mean ± standard deviation)
Hidden state estimates
Probability of regime changes (anomalies) from Switching Kalman filter (SKF)

Plot train, validation, and test data with optional standardization and NaN filtering.

Parameters:

data_processor (DataProcess) – Data processing object.
standardization (bool, optional) – Plot data in standardized or original space
plot_train_data (bool, optional) – If True, plot training data.
plot_validation_data (bool, optional) – If True, plot validation data.
plot_test_data (bool, optional) – If True, plot test data.
plot_column (list[int], optional) – List of column indices to plot.
sub_plot (plt.Axes, optional) – Matplotlib subplot axis to plot on.
color (str, optional) – Line color for plot lines.
linestyle (str, optional) – Line style (e.g., ‘-’, ‘–‘).
train_label (str, optional) – Legend label for training data.
validation_label (str, optional) – Legend label for validation data.
test_label (str, optional) – Legend label for test data.
plot_nan (bool, optional) – Whether to include NaNs in the plot for plotting missing values.

Examples

>>> from canari import DataProcess, plot_data
>>> # Create data
>>> dt_index = pd.date_range(start="2025-01-01", periods=11, freq="H")
>>> data = pd.DataFrame({'value': np.linspace(0.1, 1.0, 11)},
                index=dt_index)
>>> dp = DataProcess(data,
            train_split=0.7,
            validation_split=0.2,
            test_split=0.1,
            time_covariates = ["hour_of_day"],
            standardization=True)
>>> fig, ax = plt.subplots(figsize=(12, 4))
>>> plot_data(
        data_processor=dp,
        standardization=True,
        plot_validation_data=True,
        plot_test_data=True,
    )

Plot predicted mean and uncertainty for each data split. The uncertainty is represented by confidence regions based on a multiple of the associated variable’s standard deviation.

Parameters:

data_processor (DataProcess) – Data processing object.
mean_train_pred (np.ndarray, optional) – Predicted means for training.
std_train_pred (np.ndarray, optional) – Standard deviations for training predictions.
mean_validation_pred (np.ndarray, optional) – Predicted means for validation.
std_validation_pred (np.ndarray, optional) – Standard deviations for validation predictions.
mean_test_pred (np.ndarray, optional) – Predicted means for test predictions.
std_test_pred (np.ndarray, optional) – Standard deviations for test predictions.
num_std (int, optional) – Number of std deviations for confidence region.
sub_plot (plt.Axes, optional) – Matplotlib subplot axis to plot on.
color (str, optional) – Line color.
linestyle (str, optional) – Line style.
train_label (List[str], optional) – [mean_label, std_label] for train.
validation_label (List[str], optional) – [mean_label, std_label] for validation.
test_label (List[str], optional) – [mean_label, std_label] for test.

Examples

>>> from canari import plot_prediction
>>> mu_preds_val, std_preds_val, states = model.lstm_train(train_data=train_set,validation_data=val_set)
>>> plot_prediction(
        data_processor=dp,
        mean_validation_pred=mu_preds_val,
        std_validation_pred=std_preds_val,
    )

canari.data_visualization.plot_skf_states(data_processor: DataProcess, states: StatesHistory, model_prob: ndarray, states_to_plot: list[str] | None = 'all', states_type: str | None = 'posterior', standardization: bool | None = False, num_std: int | None = 1, plot_observation: bool | None = True, color: str | None = 'b', linestyle: str | None = '-', legend_location: str | None = None, plot_nan: bool | None = True)[source]#

Plot hidden states along with probabilities of regime changes.

Parameters:

data_processor (DataProcess) – Data processing object.
states (StatesHistory) – Object containing hidden states history over time.
model_prob (np.ndarray) – Probabilities of the abnormal regime.
states_to_plot (list[str] or "all", optional) – Names of states to plot.
states_type (str, optional) – Type of state (‘posterior’ or ‘prior’ or ‘smooth’).
standardization (bool, optional) – Whether to plot hidden states in standardized or original space.
num_std (int, optional) – Number of standard deviations for confidence regions.
plot_observation (bool, optional) – Whether to include observed data into “level” plot.
color (str, optional) – Line color for states.
linestyle (str, optional) – Line style.
legend_location (str, optional) – Location of legend in top subplot.
plot_nan (bool, optional) – Whether to include NaNs in the plot for plotting missing values.

Examples

>>> from canari import plot_skf_states
>>> filter_marginal_abnorm_prob, states = skf.filter(data=all_data)
>>> fig, ax = plot_skf_states(
        data_processor=dp,
        states=states,
        states_type="posterior",
        model_prob=filter_marginal_abnorm_prob,
    )

Plot hidden states with mean and confidence regions.

Parameters:

data_processor (DataProcess) – Data processing object.
states (StatesHistory) – Object containing hidden states history over time.
states_to_plot (list[str] or "all", optional) – Names of states to plot.
states_type (str, optional) – Type of state (‘posterior’ or ‘prior’ or ‘smooth’).
standardization (bool, optional) – Whether to plot hidden states in standardized or original space.
num_std (int, optional) – Number of standard deviations for confidence region.
sub_plot (plt.Axes, optional) – Matplotlib subplot axis to plot on.
color (str, optional) – Color for mean line and confidence region fill.
linestyle (str, optional) – Line style.
legend_location (str, optional) – Legend placement for first state subplot.

Examples

>>> from canari import plot_states
>>> mu_preds_val, std_preds_val, states = model.lstm_train(train_data=train_set,validation_data=val_set)
>>> fig, ax = plot_states(
        data_processor=dp,
        states=states,
        states_type="posterior"
    )

Plot mean and confidence region for a variable.

Parameters:

time (np.ndarray) – Time vector.
mu (np.ndarray) – Mean values to plot.
std (np.ndarray) – Standard deviation values.
color (str, optional) – Line/fill color.
linestyle (str, optional) – Line style.
label (List[str], optional) – Labels for mean and confidence region.
index (int, optional) – Index if plotting from a multivariate tensor.
num_std (int, optional) – Number of standard deviations for confidence regions.
ax (plt.Axes, optional) – Axis to use for plotting.

canari.data_visualization

Contents

canari.data_visualization#