"""Predicted vs Actual plot for regression model evaluation."""
from typing import Any, Literal, Optional, Tuple
import numpy as np
from matplotlib.axes import Axes
from matplotlib.figure import Figure
from chemotools.plotting._base import BasePlot, ColoringMixin
from chemotools.plotting._utils import scatter_with_colormap, validate_data
[ドキュメント]
class PredictedVsActualPlot(BasePlot, ColoringMixin):
"""Scatter plot of predicted vs actual values to assess regression fit.
This class creates scatter plots comparing predicted values against actual
(true) values with an ideal prediction line (y=x). Useful for visually
assessing model accuracy and detecting systematic errors or bias.
Parameters
----------
y_true : np.ndarray
True (actual) y values with shape (n_samples,) for univariate or
(n_samples, n_targets) for multivariate regression.
y_pred : np.ndarray
Predicted y values with same shape as y_true.
target_index : int, optional
For multivariate predictions, which target to plot (default: 0).
Ignored if y_true/y_pred are 1D.
color_by : np.ndarray, optional
Values for coloring samples. Can be either:
- Continuous (numeric): shows colorbar
- Categorical (strings/classes): shows legend with discrete colors
label : str, optional
Legend label for this dataset (default: None).
color : str, optional
Color for all points when color_by is None (default: auto-assigned).
colormap : str, optional
Colormap name. Colorblind-friendly defaults:
- "tab10" for categorical data
- "viridis" for continuous data
marker : str, optional
Marker style for scatter points (default: "o").
Examples: "o", "s", "^", "v", "D".
add_ideal_line : bool, optional
Whether to add diagonal y=x line showing ideal predictions (default: True).
color_mode : {"continuous", "categorical"}, optional
Explicitly specify coloring mode. If None (default), automatically
detects based on dtype and unique values of color_by.
colorbar_label : str, optional
Label for the colorbar when using continuous coloring.
Default is "Value". Only applies when color_by is continuous.
Raises
------
ValueError
If y_true and y_pred have mismatched shapes.
Examples
--------
**Basic predicted vs actual plot:**
>>> plot = PredictedVsActualPlot(y_true, y_pred)
>>> fig = plot.show(title="Model Performance")
**With categorical coloring (e.g., by batch):**
>>> batches = np.array(['A', 'B', 'A', 'B', ...])
>>> plot = PredictedVsActualPlot(y_true, y_pred, color_by=batches)
>>> fig = plot.show(title="Predictions by Batch")
**Multiple models compared:**
>>> fig, ax = plt.subplots()
>>> PredictedVsActualPlot(
... y_true, y_pred_model1, label="Model 1", color="blue"
... ).render(ax)
>>> PredictedVsActualPlot(
... y_true, y_pred_model2, label="Model 2", color="red"
... ).render(ax)
>>> ax.legend()
>>> plt.show()
**Multivariate regression:**
>>> fig, axes = plt.subplots(1, 3, figsize=(15, 5))
>>> for i in range(3):
... PredictedVsActualPlot(y_true, y_pred, target_index=i).render(axes[i])
... axes[i].set_title(f"Target {i+1}")
>>> plt.tight_layout()
>>> plt.show()
"""
def __init__(
self,
y_true: np.ndarray,
y_pred: np.ndarray,
*,
target_index: int = 0,
color_by: Optional[np.ndarray] = None,
label: Optional[str] = None,
color: Optional[str] = None,
colormap: Optional[str] = None,
marker: str = "o",
add_ideal_line: bool = True,
color_mode: Optional[Literal["continuous", "categorical"]] = None,
colorbar_label: str = "Value",
):
# Validate and store input data
self.y_true = validate_data(y_true, name="y_true", ensure_2d=False)
self.y_pred = validate_data(y_pred, name="y_pred", ensure_2d=False)
self.target_index = target_index
self.label = label
self.color = color
self.marker = marker
self.add_ideal_line = add_ideal_line
self.colorbar_label = colorbar_label
# Validate shapes match
if self.y_true.shape != self.y_pred.shape:
raise ValueError(
f"y_true and y_pred must have same shape. "
f"Got y_true: {self.y_true.shape}, y_pred: {self.y_pred.shape}"
)
# Validate color_by if provided
if color_by is not None:
color_by = validate_data(
color_by, name="color_by", ensure_2d=False, numeric=False
)
# Extract the specific target if multivariate
if self.y_true.ndim == 2:
if target_index >= self.y_true.shape[1]:
raise ValueError(
f"target_index {target_index} is out of bounds for "
f"y_true with {self.y_true.shape[1]} targets"
)
self.y_true_1d = self.y_true[:, target_index]
self.y_pred_1d = self.y_pred[:, target_index]
else:
self.y_true_1d = self.y_true
self.y_pred_1d = self.y_pred
# Initialize coloring
self._init_coloring(
color_by,
colormap,
color_mode=color_mode,
colorbar_label=self.colorbar_label,
)
def _get_default_labels(self) -> dict[str, str]:
if self.y_true.ndim == 2:
default_title = f"Predicted vs Actual (Target {self.target_index + 1})"
else:
default_title = "Predicted vs Actual"
return {
"xlabel": "Actual",
"ylabel": "Predicted",
"title": default_title,
}
[ドキュメント]
def show(
self,
*,
figsize: Optional[Tuple[float, float]] = None,
title: Optional[str] = None,
xlabel: Optional[str] = None,
ylabel: Optional[str] = None,
xlim: Optional[Tuple[float, float]] = None,
ylim: Optional[Tuple[float, float]] = None,
**kwargs: Any,
) -> Figure:
"""Create and return a complete figure with the predicted vs axtual plot.
This method handles figure creation and then delegates to `render()`.
Parameters
----------
figsize : tuple[float, float], optional
Figure size in inches (width, height).
title : str, optional
Figure title.
xlabel : str, optional
Custom x-axis label. If None, uses existing label or default.
ylabel : str, optional
Custom y-axis label. If None, uses existing label or default.
xlim : tuple[float, float], optional
X-axis limits as (xmin, xmax).
ylim : tuple[float, float], optional
Y-axis limits as (ymin, ymax).
**kwargs : Any
Additional keyword arguments passed to the render() method.
Returns
-------
Figure
The matplotlib Figure object containing the plot.
"""
return super().show(
figsize=figsize,
title=title,
xlabel=xlabel,
ylabel=ylabel,
xlim=xlim,
ylim=ylim,
**kwargs,
)
[ドキュメント]
def render(
self,
ax: Optional[Axes] = None,
*,
xlabel: Optional[str] = None,
ylabel: Optional[str] = None,
xlim: Optional[tuple[float, float]] = None,
ylim: Optional[tuple[float, float]] = None,
**kwargs: Any,
) -> tuple[Figure, Axes]:
"""Render the plot on existing or new axes.
Parameters
----------
ax : Axes, optional
Matplotlib axes to render on. If None, creates new figure/axes.
xlabel : str, optional
Custom x-axis label. If None, uses existing label or default.
ylabel : str, optional
Custom y-axis label. If None, uses existing label or default.
xlim : tuple[float, float], optional
X-axis limits (min, max).
ylim : tuple[float, float], optional
Y-axis limits (min, max).
**kwargs : Any
Additional keyword arguments passed to scatter plot.
Returns
-------
tuple[Figure, Axes]
The Figure and Axes objects containing the plot.
"""
# Remove figsize from kwargs as it is not used in render
kwargs.pop("figsize", None)
fig, ax = super().render(
ax=ax,
xlabel=xlabel,
ylabel=ylabel,
xlim=xlim,
ylim=ylim,
**kwargs,
)
# Add colorbar for continuous data
self._add_colorbar_if_needed(ax)
# Add legend if categorical or if label is provided
if self.is_categorical or self.label:
# Only add legend if there are labeled artists
handles, labels = ax.get_legend_handles_labels()
if handles:
ax.legend()
return fig, ax
def _render_plot(self, ax: Axes, **kwargs: Any) -> None:
"""Internal method to render the plot on given axes."""
alpha = kwargs.pop("alpha", 0.7)
s = kwargs.pop("s", 50)
edgecolors = kwargs.pop("edgecolors", "none")
# Determine colors
scatter_with_colormap(
ax,
self.y_true_1d,
self.y_pred_1d,
color_by=self.color_by,
is_categorical=self.is_categorical,
colormap=self.colormap,
color=self.color if self.color is not None else "steelblue",
label=self.label,
alpha=alpha,
s=s,
marker=self.marker,
edgecolors=edgecolors,
**kwargs,
)
# Add ideal prediction line (y=x)
if self.add_ideal_line:
lims = [
min(self.y_true_1d.min(), self.y_pred_1d.min()),
max(self.y_true_1d.max(), self.y_pred_1d.max()),
]
ax.plot(
lims, lims, "k--", alpha=0.5, zorder=0, label="Ideal", linewidth=1.5
)
