ml4cps.tools module

Various methods to transform data.

ml4cps.tools.binary_ordinal_encode(column, order)

Encodes a pandas Series with binary ordinal encoding based on the specified order.

Parameters:

• column (pd.Series) – The column to encode.

• order (list) – The ordered list of unique values in the column.

Returns:

The binary ordinal encoded DataFrame for the given column.

Return type:

pd.DataFrame

ml4cps.tools.composite_f1_score(anom_labels, start_event_idx, true_anom_idx)

ml4cps.tools.compute_purity(cluster_assignments, class_assignments)

ml4cps.tools.create_events_from_concurent_logs(data)

ml4cps.tools.create_events_from_signal_vectors(data, sig_names=None)

ml4cps.tools.data_list_to_dataframe(element, data, signal_names, prefix=None, last_var=None)

ml4cps.tools.df_to_torch_tensor(df)

ml4cps.tools.dict_to_csv(d, name='csv.csv')

ml4cps.tools.dict_to_df(d)

ml4cps.tools.encode_columns_to_string(df, name='Mode')

ml4cps.tools.encode_nominal(x, columns=None, categories=None)

ml4cps.tools.encode_nominal_list_df(dfs, columns=None, categories=None)

ml4cps.tools.encode_ordinal(x, columns, order=None)

ml4cps.tools.extend_derivative(signals, use_derivatives=(0, 1))

ml4cps.tools.filter_na_and_constant(data)

ml4cps.tools.filter_signals(data, sig_names)

ml4cps.tools.flatten_dict(dict_of_lists)

ml4cps.tools.flatten_dict_data(stateflow, reduce_keys_if_possible=True)

ml4cps.tools.generate_random_walk(start_values, steps=100)

Generates a random walk process for multiple variables.

Parameters: - start_values (list): A list of starting values for each variable. - steps (int): Number of steps in the random walk.

Returns: - pd.DataFrame: DataFrame containing the random walk process for each variable.

ml4cps.tools.get_binary_cols(df)

ml4cps.tools.group_components(comp, *states)

ml4cps.tools.group_data_on_discrete_state(data, state_column, reset_time=False, time_col=None)

ml4cps.tools.interpolate(time, state, new_time)

ml4cps.tools.melt_dataframe(df, timestamp=None)

ml4cps.tools.random_split_torch(split, *args)

ml4cps.tools.remove_timestamps_without_change(data, sig_names=None)

Removes timestamps where no values changed in comparison to the previous timestamp.

ml4cps.tools.signal_vector_to_event(previous_vec, sig_vec)

ml4cps.tools.signal_vector_to_state(sig_vec)

ml4cps.tools.split_data_on_signal_value(data, sig_name, new_value)

ml4cps.tools.split_train_valid(time, data, *other, split=0.8)

ml4cps.tools.standardize(x, mean=None, std=None)

ml4cps.tools.vec2str(discrete_data)

ml4cps.tools.window(x, window_size, window_step)