Working with plot outputs (figures & fitted models)

  run this example as a notebook — or grab the .ipynb (also linked at the bottom of the page)

hyp.plot returns a plain matplotlib (or plotly) Figure – there is no special container object to learn. Anything you can do with a Figure (fig.savefig(...), grabbing fig.axes[0] to tweak the plot, embedding it in a larger layout, etc.) just works.

If you also want access to the analyzed data and the fitted reduce/align/cluster models, pass return_model=True. Instead of the bare figure, hyp.plot then returns a dict bundle with six keys: {'fig': ..., 'xform_data': ..., 'animation': ..., 'models': ..., 'pipeline': ..., 'predict': ...}, where xform_data is the normalized/reduced/aligned data that was actually plotted, animation is the matplotlib.animation.Animation handle when animate=True (None otherwise, and for plotly figures), models records the reduce/align/cluster specs used to produce it, pipeline is the fitted hyp.Pipeline (feed it back in via pipeline= to replay the exact fitted stages on new data), and predict holds the fitted forecaster and forecasts when predict= was used (None otherwise).

Note that hyp.load returns raw data directly (e.g. a list of arrays) – there is nothing further to unpack.

• 
• 

/home/docs/checkouts/readthedocs.org/user_builds/hypertools/checkouts/latest/hypertools/plot/backend.py:1353: UserWarning: Failed to switch to any interactive backend (TkAgg, QtAgg, Qt5Agg, Qt4Agg, GTK4Agg, GTK3Agg, WXAgg). Falling back to 'Agg'.
  warnings.warn(BACKEND_WARNING)
axes type: Axes3D
/home/docs/checkouts/readthedocs.org/user_builds/hypertools/checkouts/latest/hypertools/plot/backend.py:1353: UserWarning: Failed to switch to any interactive backend (TkAgg, QtAgg, Qt5Agg, Qt4Agg, GTK4Agg, GTK3Agg, WXAgg). Falling back to 'Agg'.
  warnings.warn(BACKEND_WARNING)
bundle keys: ['animation', 'fig', 'models', 'pipeline', 'predict', 'xform_data']
number of arrays returned: 2
reduced shape (first array): (1000, 3)
reduce model spec: {'model': 'PCA', 'params': {'n_components': 3}}

# Code source: Contextual Dynamics Lab
# License: MIT

# import
import os
import tempfile
import hypertools as hyp

# load some data -- a list of arrays, ready to plot as-is
data = hyp.load('spiral')

# plot: the return value is just a matplotlib Figure
fig = hyp.plot(data, ndims=3)

# treat it like any other Figure
png_path = os.path.join(tempfile.mkdtemp(), 'spiral.png')
fig.savefig(png_path)
ax = fig.axes[0]
print(f"axes type: {type(ax).__name__}")

# ask for the fitted models and the analyzed data alongside the figure
out = hyp.plot(data, ndims=3, reduce='PCA', return_model=True)

fig2 = out['fig']
xform_data = out['xform_data']
models = out['models']

print(f"bundle keys: {sorted(out.keys())}")
print(f"number of arrays returned: {len(xform_data)}")
print(f"reduced shape (first array): {xform_data[0].shape}")
print(f"reduce model spec: {models['reduce']}")