Matplotlib等高线图绘制求助：棒球好坏球数据可视化需求

Hey there! I get you—contour plots can be tricky when you start with scattered data instead of a pre-gridded dataset. Let's walk through exactly how to turn your baseball strike/ball data into a meaningful contour plot step by step.

First, let's clarify the core issue: plt.contour requires 2D grid arrays for x, y, and z, but your data is a list of scattered (x,y) points with binary z-values (0 for strikes, 1 for balls). So we need to convert your raw data into this grid format first. Depending on what you want your contour plot to show, there are two solid approaches:

Option 1: Contours of Ball (z=1) Density (Most Statistically Meaningful) #

Since your z-values are categorical, showing the density of balls in different regions will give you a much more intuitive contour plot than trying to interpolate 0/1 values directly. We'll use kernel density estimation (KDE) for this:

import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import kde

# Replace these with your actual data arrays
x = np.array([1.1, 0.9, 2.6, 3.1, 0.3])
y = np.array([2.1, 3.2, 4.1, 1.1, 0.9])
z = np.array([0, 1, 0, 1, 1])

# Isolate the coordinates for balls (z=1)
ball_x = x[z == 1]
ball_y = y[z == 1]

# Create a regular grid to map our density onto
xi, yi = np.meshgrid(
    np.linspace(x.min(), x.max(), 100),  # 100 points along x-axis
    np.linspace(y.min(), y.max(), 100)   # 100 points along y-axis
)

# Calculate the KDE for ball positions
kernel = kde.gaussian_kde([ball_x, ball_y])
zi = kernel(np.vstack([xi.flatten(), yi.flatten()])).reshape(xi.shape)

# Plot the contours + original data
plt.figure(figsize=(8, 6))
# Draw contour lines with labels
contour_lines = plt.contour(xi, yi, zi, levels=5, cmap="Reds")
plt.clabel(contour_lines, inline=True, fontsize=10)

# Overlay original strike/ball points for context
plt.scatter(x[z == 0], y[z == 0], c="blue", label="Strike (z=0)", alpha=0.6)
plt.scatter(x[z == 1], y[z == 1], c="red", label="Ball (z=1)", alpha=0.6)

plt.xlabel("X Coordinate")
plt.ylabel("Y Coordinate")
plt.title("Contour Plot of Ball Density")
plt.legend()
plt.show()

Option 2: Contour of Strike/Ball Boundary #

If you specifically want to show the boundary between strike and ball regions, you can interpolate your binary z-values onto a grid and plot the 0.5 contour (the midpoint between 0 and 1):

import numpy as np
import matplotlib.pyplot as plt
from scipy.interpolate import griddata

# Same data setup as before
x = np.array([1.1, 0.9, 2.6, 3.1, 0.3])
y = np.array([2.1, 3.2, 4.1, 1.1, 0.9])
z = np.array([0, 1, 0, 1, 1])

# Create the grid
xi, yi = np.meshgrid(
    np.linspace(x.min(), x.max(), 100),
    np.linspace(y.min(), y.max(), 100)
)

# Interpolate z-values onto the grid (use 'nearest' for sharp boundaries, 'linear' for smoother)
zi = griddata((x, y), z, (xi, yi), method="nearest")

# Plot the boundary contour + filled regions
plt.figure(figsize=(8, 6))
# Draw the 0.5 boundary line
plt.contour(xi, yi, zi, levels=[0.5], colors="black", linewidths=2)
# Fill regions with color based on z-value
plt.contourf(xi, yi, zi, cmap="coolwarm", alpha=0.3)

# Overlay original points
plt.scatter(x[z == 0], y[z == 0], c="blue", label="Strike (z=0)", alpha=0.6)
plt.scatter(x[z == 1], y[z == 1], c="red", label="Ball (z=1)", alpha=0.6)

plt.xlabel("X Coordinate")
plt.ylabel("Y Coordinate")
plt.title("Strike/Ball Boundary Contour Plot")
plt.legend()
plt.colorbar(label="Z Value (0=Strike, 1=Ball)")
plt.show()

Key Notes: #

• The linspace(..., 100) parameter controls the resolution of your grid—higher numbers mean smoother contours but more computation.
• For KDE, you can adjust the bandwidth of the kernel with kernel.set_bandwidth(kernel.factor * 0.5) (smaller values = sharper density peaks) if needed.
• If your dataset is very large, consider downsampling it first to speed up interpolation/KDE calculations.

内容的提问来源于stack exchange，提问作者JohnB