energy_fault_detector.utils.visualisation

Visualisation functions.

animate_bias(bias_list, selected_column_names, filename='arcana_bias.gif', figsize=(8, 8))[source]

Plots biases as a bar plot and animates it over the iterations.

Parameters:

  • bias_list (List[DataFrame]) – A list of pandas DataFrames containing ARCANA biases recorded over the iterations.

  • selected_column_names (List[str]) – Names of the features which should be plotted.

  • filename (str) – Name of the GIF file. Defaults to ‘arcana_bias.gif’.

  • figsize (Tuple) – Size of the figure if a new one is created. Defaults to (8, 8).

animate_corrected_input(corrected_list, selected_column_names, initial_input, expected_result=None, filename='arcana_corrected_input.gif', figsize=(8, 8))[source]

Plots graphs of ARCANA corrected inputs, which are the initial_input + bias, and animates them.

Parameters:

  • corrected_list (List[DataFrame]) – A list of pandas DataFrames containing ARCANA corrected inputs recorded over the iterations.

  • selected_column_names (List[str]) – Names of the features which should be plotted.

  • initial_input (DataFrame) – Input at the starting point of the ARCANA optimization.

  • expected_result (DataFrame) – Expected normal behavior for debugging and verifying results. Defaults to None.

  • filename (str) – Name of the GIF file. Defaults to ‘arcana_corrected_input.gif’.

  • figsize (Tuple) – Size of the figure if a new one is created. Defaults to (8, 8).

plot_arcana_importance_series(importances, num_features=5, anomaly_events=None, ax=None, figsize=(8, 3), **subplot_kwargs)[source]

Plots the importance time series for the features with the highest average importance across all timestamps.

Parameters:

  • importances (List[DataFrame]) – Contains pandas DataFrames with importance values for different features (columns) across a range of timestamps (index).

  • num_features (Optional[int]) – Number of features to plot for each event. The features with the highest maximum importance across all analyzed event timestamps are selected. Defaults to 5.

  • anomaly_events (Optional[DataFrame]) – DataFrame with columns ‘start’, ‘end’, and ‘duration’ describing detected anomaly events. Defaults to None.

  • figsize (Optional[Tuple[float, float]]) – Size of the figure if a new one is created. Defaults to (8, 8).

  • ax (Optional[Axes]) – Axes object to plot on. If None, a new figure and axes will be created.

  • **subplot_kwargs – Additional keyword arguments for plt.subplots().

Returns:

None

plot_arcana_losses(losses, figsize=(8, 8))[source]

Plots the graphs for the ARCANA losses: Loss 1, Loss 2, and Combined Loss.

Parameters:

  • losses (DataFrame) – A DataFrame with iteration numbers as index and the columns representing the losses.

  • figsize (Tuple) – Size of the figure if a new one is created. Defaults to (8, 8).

Return type:

None

plot_arcana_mean_importances(importances, top_n_features=10, figsize=(8, 8), ax=None, **subplot_kwargs)[source]

Plot the ARCANA importances as a horizontal bar plot.

Parameters:

  • importances (Series) – Series containing the ARCANA importances of the features.

  • top_n_features (int) – Number of features to plot. Defaults to 10.

  • figsize (Tuple) – Size of the figure if a new one is created. Defaults to (8, 8).

  • ax (Axes) – Axes object to plot on. If None, a new figure and axes will be created.

  • **subplot_kwargs – Additional keyword arguments for plt.subplots().

Returns:

The figure and axes containing the plot for further customization if needed.

Return type:

Tuple[Figure, Axes]

plot_learning_curve(model, ax=None, label='', **subplot_kwargs)[source]

Plot the learning curve of the specified model.

Parameters:

  • model (Union[Autoencoder, FaultDetector]) – The model for which to plot the learning curve.

  • ax (Axes) – Axes to plot the learning curve on. Defaults to None.

  • label (str) – Label for the learning curve. Defaults to ‘’.

  • subplot_kwargs (dict, optional) – Additional keyword arguments for subplots, if no axes are passed.

Returns:

The figure and axes containing the plot.

Return type:

Tuple[Figure, Axes]

plot_reconstruction(data, reconstruction, features_to_plot=None, height_multiplier=1.5, original_scale=True)[source]

Plots the original dataset and its reconstruction.

Notes

  • Can result in a very large plot, if the dataset contains many columns/features. Use the

features_to_plot parameter to specify which columns to plot.

  • For MultiIndex data (e.g. multiple devices), pass data for a single group/device, i.e. df.loc[device_id] to select one device before plotting.

Parameters:

  • data (DataFrame) – DataFrame containing the original data.

  • reconstruction (DataFrame) – DataFrame containing the reconstructed data.

  • features_to_plot (Optional[List[str]]) – List of features to plot. Defaults to None (all features).

  • height_multiplier (float) – Multiplier for the vertical size of the figure. Defaults to 1.5.

  • original_scale (bool) – Whether to scale the y-axis using the input data. Defaults to True. If true, the y limits are set to minimum - std, maximum + std for each feature plotted.

Returns:

The figure and axes containing the plots, for further customization if needed.

Return type:

Tuple[Figure, Axes]

plot_reconstruction_with_model(model, data, features_to_plot=None, height_multiplier=1.5, original_scale=True)[source]
Plots the data and its reconstruction using the provided model. Similar to plot_reconstruction, but uses the

‘model.predict’ method to get the reconstruction. Counter values are plottet as diffs or rates with their reconstruction.

Note

For MultiIndex data (e.g. multiple devices), pass data for a single group/device. Use df.loc[device_id] to select one device before plotting.

Parameters:

  • model (FaultDetector) – Fitted model with data_preprocessor and autoencoder.

  • data (DataFrame) – Raw input data.

  • features_to_plot (Optional[List[str]]) – Columns to plot. If None, uses reconstruction columns.

  • height_multiplier (float) – Vertical scaling for the figure. Defaults to 1.5.

  • original_scale (bool) – If True, y-limits are based on the observed plot-series (min-std, max+std). Defaults to True.

Returns:

The figure and axes.

Return type:

Tuple[Figure, Axes]

plot_score_with_threshold(model, data, normal_index=None, ax=None, figsize=(8, 3), show_predicted_anomaly=False, show_threshold=True, show_criticality=False, max_criticality=144, score_color=None, anomaly_color=None, criticality_color='C2', marker_size=1, threshold_color='k', **subplot_kwargs)[source]

Plots the anomaly scores of the AnomalyDetector model along with the threshold for the provided data.

Note

For MultiIndex data (e.g. multiple devices), pass data for a single group/device. Use df.loc[device_id] to select one device before plotting.

Parameters:

  • model (FaultDetector) – The anomaly detection model used to compute the scores.

  • data (DataFrame) – DataFrame containing the data for which scores are computed.

  • normal_index (Series) – Boolean series indicating whether the data points have a normal status or not.

  • ax (Axes) – Axes object to plot on. If None, a new figure and axes will be created.

  • figsize (Tuple[float, float]) – Size of the figure if a new one is created. Defaults to (8, 3).

  • show_predicted_anomaly (bool) – Whether to show the predicted anomaly scores. Defaults to False.

  • show_threshold (bool) – Whether to show the threshold scores. Defaults to True.

  • show_criticality (bool) – Whether to show the criticality counter. Defaults to False.

  • max_criticality (int) – If show_criticality is True, the maximum value of the criticality counter can be specified. Defaults to 144 (one day of 10 min timestamps).

  • score_color (Optional[str]) – Color to use for the anomaly score.

  • anomaly_color (Optional[str]) – Color to use for the anomalous data points (using normal_index).

  • criticality_color (Optional[str]) – Color to use for the criticality counter if show_criticality is True. Defaults to ‘C2’.

  • marker_size (Optional[int]) – Dot size in the scatter plot. Defaults to 1.

  • threshold_color (Optional[str]) – Color to use for the threshold.

  • **subplot_kwargs – Additional keyword arguments for plt.subplots().

Returns:

The figure and axes containing the plot for further customization if needed.

Return type:

Tuple[Figure, Axes]