Visualization Best Practices

Edward Tufte's Data-Ink Ratio:
  Data-Ink Ratio = Data-Ink / Total Ink

  Maximize = remove everything that doesn't encode data

REMOVE:
  ❌ Gridlines that don't aid reading
  ❌ Heavy borders around the chart
  ❌ 3D effects (always)
  ❌ Unnecessary tick marks
  ❌ Background colors or gradients
  ❌ Decorative elements ("chart junk")
  ❌ Redundant legends (label directly)

KEEP:
  ✅ Data series (bars, lines, points)
  ✅ Essential gridlines (y-axis only for bar charts)
  ✅ Axis labels with units
  ✅ Clear title
  ✅ Reference lines (targets, averages)
  ✅ Annotations for key insights

import matplotlib.pyplot as plt
import numpy as np

products = ['Laptop', 'Phone', 'Monitor', 'Desk', 'Chair']
sales    = [850, 1240, 430, 620, 380]

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(14, 6))

# BAD CHART — common mistakes
ax1.bar(products, sales, color=['red','green','blue','orange','purple'], width=0.9)
ax1.set_title('Product Sales', fontsize=9)
ax1.set_xlabel('Product Name')
ax1.set_ylabel('Units')
ax1.grid(True)
for spine in ax1.spines.values():
    spine.set_linewidth(2)
ax1.set_facecolor('#EEEEEE')  # Grey background
ax1.tick_params(axis='both', length=5)
ax1.legend(['Sales'], loc='upper right')
ax1.set_title('❌ POOR: Chart junk, wrong colors, full grid', fontsize=10)

# GOOD CHART — best practices
bars = ax2.barh(products, sales, color=['#1565C0' if s == max(sales) else '#90CAF9' for s in sales],
                 height=0.6, edgecolor='white')
for bar, val in zip(bars, sales):
    ax2.text(val + 5, bar.get_y() + bar.get_height()/2,
              f'{val:,} units', va='center', fontsize=10)
ax2.set_title('✅ GOOD: Clean, sorted, direct labels', fontsize=10, fontweight='bold')
ax2.set_xlabel('Units Sold')
ax2.spines['top'].set_visible(False)
ax2.spines['right'].set_visible(False)
ax2.spines['left'].set_visible(False)
ax2.set_xticks([])   # Remove x-axis since values are labeled
ax2.grid(False)      # No gridlines needed with direct labels

plt.suptitle('Best Practices: Before vs After', fontsize=13, fontweight='bold')
plt.tight_layout()
plt.savefig('best_practices.png', dpi=150)
plt.show()

1. START BARS AT ZERO — bar length encodes value
2. USE POSITION > LENGTH > COLOR > ANGLE > SIZE for accuracy
3. SORT FOR INSIGHT — sorted bars reveal ranking immediately
4. LIMIT COLORS — max 7-8 categorical colors; use grey for context
5. LABEL DIRECTLY — data labels beat legends for small datasets
6. REMOVE CHART JUNK — eliminate decorative elements
7. ONE MESSAGE PER CHART — design for one clear takeaway
8. MOBILE-FRIENDLY — fonts ≥ 11px, enough contrast
9. CITE YOUR DATA — always include source and date
10. TELL A STORY — "what, so what, now what" structure