🖼️ Plotting

This module provides plotting functions for conformal prediction.

plotting.draw_bounding_box(box=None, label='', image=None, image_path=None, color='red', legend='', show=False)

Draw a bounding box on a given image.

Parameters:

  • box (tuple) – the coordinates of the bounding box in the format (x_min, y_min, x_max, y_max).

  • label (str) – the label for the bounding box.

  • image (PIL.Image.Image) – the image object to draw on. If provided, image_path is ignored.

  • image_path (str) – the path to the image file. Should be provided if image is not.

  • color (str) – the color of the bounding box outline and label.

  • legend (str) – the legend for the plot.

  • show (bool) – whether to display the image with the bounding box and legend.

Returns:

the image object with the bounding box and label.

Return type:

PIL.Image.Image

plotting.plot_prediction_intervals(y_true, y_pred_lower, y_pred_upper, X=None, y_pred=None, ax=None, **fig_kw)

Plot prediction intervals whose bounds are given by y_pred_lower and y_pred_upper. True and predicted (if provided) point values are also displayed.

Parameters:

  • y_true (ndarray) – true output values.

  • y_pred_lower (ndarray) – lower bounds of the prediction intervals.

  • y_pred_upper (ndarray) – upper bounds of the prediction intervals.

  • X (ndarray, optional) – abscisse vector.

  • y_pred (ndarray, optional) – predicted values.

  • ax (matplotlib.axes.Axes, optional) – plot using the provided axis. Otherwise, a new figure is created and the corresponding axis is returned as output.

  • fig_kw – all additional keyword arguments are passed to the pyplot.figure call.

Returns:

updated axis if ax provided. Otherwise a new figure is created and the corresponding axis is returned as output

Return type:

matplotlib.axes.Axes

Example 1:

import numpy as np
from sklearn.datasets import make_regression
from deel.puncc.plotting import plot_prediction_intervals

np.random.seed(0)

# Generate a random regression problem
X, y = make_regression(n_samples=100, n_features=100, n_informative=20,
                random_state=0, shuffle=False)

# Generate dummy point predictions
y_pred = y + np.random.normal(loc=0.0, scale=100, size=len(y))
# Generate dummy interval predictions
y_pred_lower = y_pred - 120 - np.random.normal(loc=0.0, scale=20.0, size=len(y))
y_pred_upper = y_pred + 120 + np.random.normal(loc=0.0, scale=20.0, size=len(y))

# Plot the prediction interval. The optional argument "y_pred" is
# provided and the prediction is displayed.
# Additionnaly, two keyword arguments "figsize" and "loc" are given to
# configure the figure.
ax = plot_prediction_intervals(
y_true=y,
y_pred=y_pred,
y_pred_lower=y_pred_lower,
y_pred_upper=y_pred_upper,
figsize=(20,10),
loc="best")

Example 2:

import numpy as np
import matplotlib.pyplot as plt
from sklearn.datasets import make_regression
from deel.puncc.plotting import plot_prediction_intervals

np.random.seed(0)

# Generate a random regression problem
X, y = make_regression(n_samples=100, n_features=1, noise=20,
                random_state=0, shuffle=False)

# Generate dummy point predictions
y_pred = y + np.random.normal(loc=0.0, scale=10, size=len(y))

# Generate two dummy interval predictions
## First
y_pred_lower1 = y_pred - 15
y_pred_upper1 = y_pred + 15
# Second
y_pred_lower2 = y_pred - 10 - np.random.normal(loc=0.0, scale=2.0, size=len(y))
y_pred_upper2 = y_pred + 10 + np.random.normal(loc=0.0, scale=2.0, size=len(y))

# Create a grid of figures
fig, ax = plt.subplots(nrows=2, figsize=(20,10))

# Plot first figure by providing the axis "ax[0]"
plot_prediction_intervals(
X = X[:,0],
y_true=y,
y_pred=y_pred,
y_pred_lower=y_pred_lower1,
y_pred_upper=y_pred_upper1,
ax=ax[0])

# Plot second figure by providing the axis "ax[1]"
plot_prediction_intervals(
X = X[:,0],
y_true=y,
y_pred=y_pred,
y_pred_lower=y_pred_lower2,
y_pred_upper=y_pred_upper2,
ax=ax[1])