Colormap Selection¶

Choosing the right colormap is critical for effective data visualization. This guide covers colormap categories, selection criteria, and best practices.

Colormap Categories¶

Matplotlib provides several categories of colormaps:

Setup¶

```python import matplotlib.pyplot as plt import numpy as np

x = np.linspace(-3, 3, 100) y = np.linspace(-3, 3, 100) X, Y = np.meshgrid(x, y) ```

Sequential Colormaps¶

Best for data that progresses from low to high values.

Perceptually Uniform¶

These colormaps have uniform perceptual brightness changes, making them ideal for scientific visualization.

```python Z = np.exp(-(X2 + Y2))

fig, axes = plt.subplots(2, 3, figsize=(15, 9))

cmaps = ['viridis', 'plasma', 'inferno', 'magma', 'cividis', 'turbo']

for ax, cmap in zip(axes.flat, cmaps): im = ax.imshow(Z, cmap=cmap, origin='lower') ax.set_title(f"'{cmap}'") ax.axis('off') plt.colorbar(im, ax=ax, shrink=0.8)

plt.suptitle('Perceptually Uniform Sequential Colormaps', fontsize=14) plt.tight_layout() plt.show() ```

Single-Hue Sequential¶

```python fig, axes = plt.subplots(2, 3, figsize=(15, 9))

cmaps = ['Blues', 'Greens', 'Reds', 'Purples', 'Oranges', 'Greys']

for ax, cmap in zip(axes.flat, cmaps): im = ax.imshow(Z, cmap=cmap, origin='lower') ax.set_title(f"'{cmap}'") ax.axis('off') plt.colorbar(im, ax=ax, shrink=0.8)

plt.suptitle('Single-Hue Sequential Colormaps', fontsize=14) plt.tight_layout() plt.show() ```

Multi-Hue Sequential¶

```python fig, axes = plt.subplots(2, 3, figsize=(15, 9))

cmaps = ['YlOrRd', 'YlGnBu', 'BuPu', 'GnBu', 'PuBuGn', 'OrRd']

for ax, cmap in zip(axes.flat, cmaps): im = ax.imshow(Z, cmap=cmap, origin='lower') ax.set_title(f"'{cmap}'") ax.axis('off') plt.colorbar(im, ax=ax, shrink=0.8)

plt.suptitle('Multi-Hue Sequential Colormaps', fontsize=14) plt.tight_layout() plt.show() ```

When to Use Sequential¶

• Density plots
• Probability distributions
• Temperature (single direction)
• Elevation maps
• Any monotonically increasing data

```python

Example: Probability density¶

from scipy import stats

rv = stats.multivariate_normal([0, 0], [[1, 0], [0, 1]]) Z_pdf = rv.pdf(np.dstack((X, Y)))

fig, ax = plt.subplots(figsize=(8, 6)) im = ax.imshow(Z_pdf, extent=[-3, 3, -3, 3], origin='lower', cmap='viridis') ax.set_title('Bivariate Normal PDF') ax.set_xlabel('x') ax.set_ylabel('y') plt.colorbar(im, ax=ax, label='Density') plt.show() ```

Diverging Colormaps¶

Best for data with a meaningful center point (often zero).

Standard Diverging¶

```python Z_div = X2 - Y2 # Saddle function with positive and negative values

fig, axes = plt.subplots(2, 3, figsize=(15, 9))

cmaps = ['coolwarm', 'RdBu', 'seismic', 'bwr', 'RdYlBu', 'PiYG']

for ax, cmap in zip(axes.flat, cmaps): im = ax.imshow(Z_div, cmap=cmap, origin='lower') ax.set_title(f"'{cmap}'") ax.axis('off') plt.colorbar(im, ax=ax, shrink=0.8)

plt.suptitle('Diverging Colormaps', fontsize=14) plt.tight_layout() plt.show() ```

Centered Normalization¶

For diverging data, center the colormap at zero:

```python from matplotlib.colors import TwoSlopeNorm

Z_asym = X2 - Y2 + 2 # Asymmetric around zero

fig, axes = plt.subplots(1, 2, figsize=(12, 5))

Without centering¶

im1 = axes[0].imshow(Z_asym, cmap='RdBu', origin='lower') axes[0].set_title('Default Normalization') plt.colorbar(im1, ax=axes[0])

With centering at zero¶

norm = TwoSlopeNorm(vmin=Z_asym.min(), vcenter=0, vmax=Z_asym.max()) im2 = axes[1].imshow(Z_asym, cmap='RdBu', norm=norm, origin='lower') axes[1].set_title('Centered at Zero (TwoSlopeNorm)') plt.colorbar(im2, ax=axes[1])

plt.tight_layout() plt.show() ```

When to Use Diverging¶

• Correlation matrices
• Temperature anomalies
• Profit/loss data
• Deviations from mean
• Any data with positive and negative values around a center

```python

Example: Correlation matrix¶

np.random.seed(42) data = np.random.randn(5, 100) corr = np.corrcoef(data)

fig, ax = plt.subplots(figsize=(8, 6)) im = ax.imshow(corr, cmap='RdBu', vmin=-1, vmax=1) ax.set_title('Correlation Matrix') ax.set_xticks(range(5)) ax.set_yticks(range(5)) plt.colorbar(im, ax=ax, label='Correlation') plt.show() ```

Cyclic Colormaps¶

Best for periodic data like angles or phases.

Available Cyclic Colormaps¶

```python theta = np.arctan2(Y, X) # Angle in radians

fig, axes = plt.subplots(1, 3, figsize=(15, 4))

cmaps = ['twilight', 'twilight_shifted', 'hsv']

for ax, cmap in zip(axes, cmaps): im = ax.imshow(theta, cmap=cmap, origin='lower') ax.set_title(f"'{cmap}'") ax.axis('off') plt.colorbar(im, ax=ax, shrink=0.8)

plt.suptitle('Cyclic Colormaps', fontsize=14) plt.tight_layout() plt.show() ```

When to Use Cyclic¶

• Angular data (directions, phases)
• Time of day (circular)
• Hue values
• Any data that wraps around

```python

Example: Phase angle¶

Z_complex = X + 1j * Y phase = np.angle(Z_complex)

fig, ax = plt.subplots(figsize=(8, 6)) im = ax.imshow(phase, extent=[-3, 3, -3, 3], origin='lower', cmap='twilight') ax.set_title('Complex Phase Angle') ax.set_xlabel('Re(z)') ax.set_ylabel('Im(z)') plt.colorbar(im, ax=ax, label='Phase (radians)') plt.show() ```

Qualitative Colormaps¶

Best for categorical data with no natural ordering.

Available Qualitative Colormaps¶

```python fig, axes = plt.subplots(2, 3, figsize=(15, 8))

cmaps = ['Set1', 'Set2', 'Set3', 'tab10', 'tab20', 'Paired']

Create categorical-like data¶

np.random.seed(42) Z_cat = np.random.randint(0, 8, (20, 20))

for ax, cmap in zip(axes.flat, cmaps): im = ax.imshow(Z_cat, cmap=cmap, origin='lower') ax.set_title(f"'{cmap}'") ax.axis('off') plt.colorbar(im, ax=ax, shrink=0.8)

plt.suptitle('Qualitative Colormaps', fontsize=14) plt.tight_layout() plt.show() ```

When to Use Qualitative¶

• Segmentation maps
• Cluster labels
• Land use categories
• Any discrete, unordered categories

```python

Example: Cluster visualization¶

from sklearn.datasets import make_blobs from sklearn.cluster import KMeans

X_data, _ = make_blobs(n_samples=300, centers=4, random_state=42) kmeans = KMeans(n_clusters=4, random_state=42).fit(X_data)

fig, ax = plt.subplots(figsize=(8, 6)) scatter = ax.scatter(X_data[:, 0], X_data[:, 1], c=kmeans.labels_, cmap='Set1') ax.set_title('K-Means Clustering') plt.colorbar(scatter, ax=ax, label='Cluster') plt.show() ```

Reversed Colormaps¶

Add _r suffix to reverse any colormap.

```python fig, axes = plt.subplots(2, 3, figsize=(15, 9))

Z = np.exp(-(X2 + Y2))

pairs = [ ('viridis', 'viridis_r'), ('Blues', 'Blues_r'), ('coolwarm', 'coolwarm_r') ]

for i, (cmap1, cmap2) in enumerate(pairs): im1 = axes[0, i].imshow(Z, cmap=cmap1, origin='lower') axes[0, i].set_title(f"'{cmap1}'") axes[0, i].axis('off') plt.colorbar(im1, ax=axes[0, i], shrink=0.8)

im2 = axes[1, i].imshow(Z, cmap=cmap2, origin='lower')
axes[1, i].set_title(f"'{cmap2}'")
axes[1, i].axis('off')
plt.colorbar(im2, ax=axes[1, i], shrink=0.8)

plt.suptitle('Original vs Reversed Colormaps', fontsize=14) plt.tight_layout() plt.show() ```

Color Vision Deficiency¶

Consider colorblind-friendly options.

Colorblind-Safe Colormaps¶

```python fig, axes = plt.subplots(2, 3, figsize=(15, 9))

Colorblind-friendly options¶

cmaps = ['viridis', 'plasma', 'cividis', 'inferno', 'magma', 'YlOrBr']

for ax, cmap in zip(axes.flat, cmaps): im = ax.imshow(Z, cmap=cmap, origin='lower') ax.set_title(f"'{cmap}' (colorblind-safe)") ax.axis('off') plt.colorbar(im, ax=ax, shrink=0.8)

plt.suptitle('Colorblind-Friendly Colormaps', fontsize=14) plt.tight_layout() plt.show() ```

Avoid These for Colorblind Users¶

• jet (rainbow) - poor perceptual uniformity
• hsv - hard to distinguish regions
• rainbow - misleading intensity perception

```python fig, axes = plt.subplots(1, 3, figsize=(15, 4))

Problematic colormaps¶

cmaps = ['jet', 'rainbow', 'hsv']

for ax, cmap in zip(axes, cmaps): im = ax.imshow(Z, cmap=cmap, origin='lower') ax.set_title(f"'{cmap}' (avoid for scientific use)") ax.axis('off') plt.colorbar(im, ax=ax, shrink=0.8)

plt.suptitle('Colormaps to Avoid', fontsize=14) plt.tight_layout() plt.show() ```

Custom Colormaps¶

Create custom colormaps for specific needs.

From Listed Colors¶

```python from matplotlib.colors import ListedColormap

colors = ['#2ecc71', '#3498db', '#9b59b6', '#e74c3c', '#f39c12'] custom_cmap = ListedColormap(colors)

fig, ax = plt.subplots(figsize=(8, 6)) im = ax.imshow(np.random.randint(0, 5, (20, 20)), cmap=custom_cmap) ax.set_title('Custom ListedColormap') plt.colorbar(im, ax=ax) plt.show() ```

Linear Segmented¶

```python from matplotlib.colors import LinearSegmentedColormap

White to blue gradient¶

colors = ['white', 'lightblue', 'blue', 'darkblue'] custom_cmap = LinearSegmentedColormap.from_list('custom_blues', colors, N=256)

fig, ax = plt.subplots(figsize=(8, 6)) im = ax.imshow(Z, cmap=custom_cmap, origin='lower') ax.set_title('Custom LinearSegmentedColormap') plt.colorbar(im, ax=ax) plt.show() ```

Truncated Colormap¶

```python def truncate_colormap(cmap, minval=0.0, maxval=1.0, n=256): new_cmap = LinearSegmentedColormap.from_list( f'trunc({cmap.name},{minval:.2f},{maxval:.2f})', cmap(np.linspace(minval, maxval, n)) ) return new_cmap

Use only middle portion of viridis¶

cmap_full = plt.cm.viridis cmap_trunc = truncate_colormap(cmap_full, 0.2, 0.8)

fig, axes = plt.subplots(1, 2, figsize=(12, 5))

im1 = axes[0].imshow(Z, cmap=cmap_full, origin='lower') axes[0].set_title('Full viridis') plt.colorbar(im1, ax=axes[0])

im2 = axes[1].imshow(Z, cmap=cmap_trunc, origin='lower') axes[1].set_title('Truncated viridis (0.2-0.8)') plt.colorbar(im2, ax=axes[1])

plt.tight_layout() plt.show() ```

Selection Decision Tree¶

Is your data... │ ├── Ordered/Sequential (low → high)? │ ├── Need colorblind-safe? → viridis, cividis, plasma │ ├── Single hue preferred? → Blues, Greens, Greys │ └── Multi-hue preferred? → YlOrRd, YlGnBu │ ├── Diverging (center is meaningful)? │ ├── Center at zero? → coolwarm, RdBu, seismic │ └── Asymmetric data? → Use TwoSlopeNorm │ ├── Cyclic (wraps around)? │ └── Use twilight or twilight_shifted │ └── Categorical (no order)? ├── Few categories (<10)? → Set1, tab10 └── Many categories? → tab20, Set3

Quick Reference Table¶

List All Available Colormaps¶

```python import matplotlib.pyplot as plt

cmaps = plt.colormaps() print(f"Total colormaps: {len(cmaps)}") print("\nFirst 20:", cmaps[:20]) ```

Visual Catalog¶

```python import matplotlib.pyplot as plt import numpy as np

gradient = np.linspace(0, 1, 256).reshape(1, -1)

categories = { 'Perceptually Uniform': ['viridis', 'plasma', 'inferno', 'magma', 'cividis'], 'Sequential': ['Blues', 'Greens', 'Reds', 'YlOrRd', 'PuBuGn'], 'Diverging': ['coolwarm', 'RdBu', 'seismic', 'bwr', 'RdYlBu'], 'Cyclic': ['twilight', 'twilight_shifted', 'hsv'], }

for category, cmap_list in categories.items(): fig, axes = plt.subplots(len(cmap_list), 1, figsize=(8, len(cmap_list) * 0.4)) fig.suptitle(category, fontsize=12, fontweight='bold')

for ax, cmap in zip(axes, cmap_list):
    ax.imshow(gradient, aspect='auto', cmap=cmap)
    ax.set_ylabel(cmap, rotation=0, ha='right', va='center', fontsize=9)
    ax.set_xticks([])
    ax.set_yticks([])

plt.tight_layout()
plt.show()

```

Exercises¶

Exercise 1. Write code that displays the same 8x8 random data array using four different colormaps in a 2x2 subplot grid. Add the colormap name as the title of each subplot.

Exercise 2. Explain the difference between sequential, diverging, and qualitative colormaps. Give one example of each and when to use it.

Exercise 3. Write code that creates a custom diverging colormap centered at zero using matplotlib.colors.TwoSlopeNorm. Apply it to data that ranges from -5 to 10.

Exercise 4. Create a figure showing how vmin and vmax affect the colormap mapping. Show the same data with default limits and with vmin=-2, vmax=2.