Utilities

Data

Finds and returns the index of the closest element to value.

Args:

array: Sorted 1D array.

value: Value to be searched.

Return type

int

tsipy.utils.data.create_dir(results_dir_path, dir_name='results')

Creates a directory with added timestamp of creation.

Return type

str

tsipy.utils.data.denormalize(y, mean, scale)

Denormalize array in the first dimension as given by x = y * scale + mean.

Return type

ndarray

tsipy.utils.data.downsample_signal(x, k=1)

Downsamples a signal uniformly with a rate of k.

Return type

ndarray

tsipy.utils.data.downsampling_indices_by_max_points(x, max_points=100000)

Computes indices of a uniformly downsampled signal of length max_points.

Return type

ndarray

tsipy.utils.data.get_time_output(t_nns, n_per_unit, min_time=None, max_time=None)

Creates a time array with n_per_unit` elements per unit.

Return type

ndarray

tsipy.utils.data.get_window_indices(x, x_start, x_end)

Obtain the start and end indices in x that are in window [x_start, x_end].

Args:

x: Sorted 1-D array. x_start: Window start. x_end: Window end.

Returns: A tuple of a start and end index of x, such that

x_start <= x[x_start_id:x_end_id + 1] <= x_end.

Return type

Tuple[int, int]

tsipy.utils.data.is_integer(num)

Checks if the input has an integer type.

Return type

bool

tsipy.utils.data.is_sorted(array)

Check if array is sorted.

Args:

array: 1-D array.

Returns:

True if array is sorted and False otherwise.

Return type

bool

tsipy.utils.data.make_dir(directory)

Creates a directory if it does not exist.

Return type

str

tsipy.utils.data.nonclipped_indices(x, n_scale=5.0)

Return non-clipped indices that are close to array mean.

Non-clipped index i satisfies:

x_mean + n_std * x_std >= x[i] >= x_mean - n_std * x_std.

Return type

ndarray

tsipy.utils.data.normalize(x, mean, scale)

Normalize array in the first dimension as given by y = (x - mean) / scale.

Return type

ndarray

tsipy.utils.data.transform_time_to_unit(t, t_label='year', start=datetime.datetime(1996, 1, 1, 0, 0))

Transforms time unit to t_label starting at start.

Examples:
>>> import numpy as np
>>> t = np.arange(0, 366, 365.25 / 4)  # Time in days
>>> transform_time_to_unit(t)  # Transformed to years
array([1996.  , 1996.25, 1996.5 , 1996.75, 1997.  ])
Return type

ndarray

Print

Pretty print utilities with indents, print blocks and colors.

tsipy.utils.print.cformat(string, color=None)

Colors the input string.

Return type

str

tsipy.utils.print.pformat(*args, shift=50, level=0, color=None)

Pretty string formatting utility function into two columns.

It formats arguments passed in two columns:

  • keyword (left aligned),

  • values (right aligned and separated by spaces).

Return type

str

tsipy.utils.print.pprint(*args, shift=50, level=0, color=None)

Pretty print utility function of arguments into two columns.

Formatting is described pformat().

Return type

None

tsipy.utils.print.pprint_block(*args, width=None, level=0, color=None)

Pretty print utility function for code sections.

Return type

None

Plot

Plot utilities for visualizing signals, correction history and signals with confidence intervals.

tsipy.utils.plot.plot_signals(signal_fourplets, results_dir=None, title=None, tight_layout=True, show=False, **kwargs)

Helper function for plotting signals.

Return type

Tuple[Figure, Axes]

tsipy.utils.plot.plot_signals_and_confidence(signal_fourplets, results_dir=None, title=None, confidence=0.95, alpha=0.5, tight_layout=False, show=False, **kwargs)

Helper function for plotting signal mean and confidence interval.

Return type

Tuple[Figure, Axes]

tsipy.utils.plot.plot_signals_history(x, signals_history, results_dir=None, title=None, n_rows=2, n_cols=2, fig_size=(12, 6), tight_layout=False, show=False, **kwargs)

Helper function for plotting degradation correction history.

Return type

Tuple[Figure, Axes]

tsipy.utils.plot.set_style(style_name='seaborn', fig_size=(12, 6), font_size=16, ax_font_size=18, ticks_font_size=16, title_font_size=16, legend_font_size=16, marker_type='x', out_format='png', latex=False)

Set pyplot style parameters.

Return type

None