Time series plots

[alondmnt]

Description

This PR adds various time series plots that are designed to be:

• Standardised
• Reusable
• Customisable
• Composable

The use cases currently focus on diet, CGM and sleep, but these functions can be adapted to application data and other time series. Some of the diet plots may be refactored to be more generalised in the future.

See also: https://github.com/PhenoAI/pheno-examples/pull/15

New features

• timeseries_plots.py
  • TimeSeriesFigure: class to handle mult-track time series plots more easily, by managing their axes together.
  • plot_events_bars: general purpose function for events.
  • plot_events_fill: general purpose function for events.
• diet_plots.py
  • plot_nutrient_bars: simple time plot of stacked bars of fields, separated according to units.
  • plot_nutrient_lollipop: fancy time plot of pie charts of nutrients.
  • plot_meals_hbars: time plot of meal content by food categories.
  • plot_diet_cgm_sleep: combines almost all of the above together.
• sleep_plots.py
  • plot_sleep: re-implemented with the above functions.

TODOs

• Add plot examples (and the corresponding data) to pheno-utils.
• Make other existing time plots compatible with TimeSeriesFigure.

Examples

3 diet plots in 1 figure

g = TimeSeriesFigure(figsize=(14, 7))

time_range = ('2021-04-05 6:00', '2021-04-06 0:00')
# Each call to the plot() methods adds a new time-synced subplot to the figure.
# A plot function only needs to have an `ax` arg to support the TimeSeriesFigure.
g.plot(plot_nutrient_lollipop, events_df, size_scale=15, time_range=time_range,
       name='diet_pie')
g.plot(plot_meals_hbars, events_df, time_range=time_range,
       name='diet_meals', height=2)
g.plot(plot_nutrient_bars, events_df, label=None, time_range=time_range,
       name='diet_bars', n_axes=3)
g.set_axis_padding(0.07)

diet X cgm X sleep

updated sleep plot