Skip to content

Latest commit

 

History

History
59 lines (33 loc) · 2.55 KB

CHANGELOG.md

File metadata and controls

59 lines (33 loc) · 2.55 KB
Raw

Changelog

0.2.0

  • First "feature-complete" release of metEAUdata. Contains functionality to transform both univariate and multivariate data.

0.2.1

  • Added the step_distance attribute to ProcessingStep. step_distanceindicates how many steps forward (or backward) a time series with this processing step is from its input time series.
  • Removed calibration_info from ProcessingStep because that info is already found in parameters.

0.3.0

  • TimeSeries, Signal and Dataset objects now have a plot method to generate basic visualizations. This adds plotly as a dependency. Other backends (namely, matplotlib) may be developed in the future.

0.3.1

  • Added input_series_names to ProcessingStep so that step sequences and dependencies can be traced.

0.4.0

  • Modified the naming convention of time series inside signals and of the signals themselves. Now, new time series are assigned a numbered name. The number is separated from the rest of the name with a # symbol. This allows multiple runs of a processing function to not overwrite the preceding versions. Signals are now also numbered, meaning that dataset processing will also not overwrite existing signals.

0.4.1

  • Signals will be assigned a number if they are unnumbered at creation.

0.4.3

  • Bug in dataset numbering
  • Added tests
  • Slight adjustments to plots

0.4.4

  • Now updating the time series names in the processing steps

0.5.0

  • Added a plottling method to Signal.plot_dependency_graph to visually represent the process that went into creating a time series from that signal.

0.5.1

  • Added a field to FunctionInfo that replicated the entire processing function's code

0.5.2

  • Fixed a bug where plots would not render if the frequency of the data was 1 second, 1 minute, 1 hour, .... Reason is that the "1" would be omitted in the series' frequency, which would throw off the plot function when trying to calculate the x coordinates of the plot.
  • Fixed a bug that overwrote the FunctionInfo.source_code field when deserializing a serialized ProcessingStep.

0.6.0

  • Fixed a typo in the processing functions where the reference pointed to the wrong GitHub repository.
  • Added a processing function (replace_ranges) that lets users replace values in a time series with a filler value (eg, NaN). Can be used to filter out manually-identified invalid data.
  • Added __str__ method to processing steps so they print nicely

0.7.0

  • Added a new univariate processing function subset that lets you slice time series to a desired length or index values