如何基于DataFrame变量组合绘制整合式散点图？

Hey Megan, I've got you covered! Let's optimize your code so you can visualize all your required scatter plot combinations automatically—no manual parameter tweaking needed. Here's a practical, clean solution:

Refactored Code for Automatic Multi-Combination Plotting #

We'll use a dictionary to map your kind labels to their DataFrames, generate all valid index pairs automatically, and use subplots to display all combinations in one organized figure.

import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
from itertools import combinations

# --- Test DataFrames ---
food = {'col1': [1, 2], 'col2': [3, 4]}
food_df = pd.DataFrame(data=food)
meat = {'col1': [3, 6], 'col2': [1, 2]}
meat_df = pd.DataFrame(data=meat)
vege = {'col1': [5, 9], 'col2': [0, 3]}
vege_df = pd.DataFrame(data=vege)

# --- Core Parameters ---
kind = ["food", "meat", "vege"]
features = ["Attack", "Volume"]
col = ["dd", "aa"]

# Map kind labels to their DataFrames (avoids messy hardcoding later)
df_mapping = {label: df for label, df in zip(kind, [food_df, meat_df, vege_df])}

# Generate all valid kind index pairs automatically (no manual (0,1) entries!)
kind_index_pairs = list(combinations(range(len(kind)), 2))

# Assign unique colors to each feature for clear differentiation
feature_colors = {"Attack": "#1f77b4", "Volume": "#ff7f0e"}

# --- Create Subplot Layout ---
# 1 row, 3 columns to fit all 3 variable combinations
fig, axes = plt.subplots(1, 3, figsize=(18, 6))

# Iterate over each kind pair and corresponding subplot
for (idx1, idx2), ax in zip(kind_index_pairs, axes.flat):
    # Grab the two DataFrames for this combination
    df_x = df_mapping[kind[idx1]]
    df_y = df_mapping[kind[idx2]]
    
    # Plot data for each feature
    for feature in features:
        # Get the row index for the current feature
        feat_row_idx = features.index(feature)
        # Extract the data points for this feature
        x_data = df_x.iloc[feat_row_idx].values
        y_data = df_y.iloc[feat_row_idx].values
        
        # Plot scatter points with unique color
        ax.scatter(x_data, y_data, color=feature_colors[feature], 
                   label=feature, s=100, alpha=0.8)
        
        # Add annotations for each column label ("dd", "aa")
        for point_idx, txt in enumerate(col):
            ax.annotate(txt, (x_data[point_idx], y_data[point_idx]),
                        xytext=(x_data[point_idx]+0.1, y_data[point_idx]+0.1),
                        fontsize=10, weight="bold")
    
    # Configure subplot labels and title
    ax.set_xlabel(kind[idx1], fontsize=12)
    ax.set_ylabel(kind[idx2], fontsize=12)
    ax.set_title(f"{kind[idx1]} vs {kind[idx2]}", fontsize=14, pad=15)
    ax.legend(title="Feature", fontsize=10)
    ax.grid(True, alpha=0.3)

# Adjust spacing between subplots for better readability
plt.tight_layout()
plt.show()

Key Improvements Breakdown: #

• Flexible Data Mapping: The df_mapping dictionary lets you quickly access any DataFrame using its kind label—adding new DataFrames later only requires updating this dictionary and the kind list.
• Automatic Pair Generation: itertools.combinations creates all valid (kind1, kind2) index pairs automatically, so you never have to manually adjust kind1/kind2 values again.
• Organized Subplot Layout: All three scatter plot combinations are displayed in one figure, making it easy to compare relationships between variables at a glance.
• Clear Visuals: Unique colors for each feature and bold annotations ensure your data points are easy to identify and interpret.
• Maintainable Code: The modular structure means you can tweak colors, point sizes, or add new features/DataFrames without rewriting large sections of code.

If you'd prefer to plot all combinations on a single axis instead of subplots, we could adjust the code to use different markers for each kind pair—but subplots are generally the clearer choice for comparing multiple variable relationships.

内容的提问来源于stack exchange，提问作者Megan