Common Visualization Mistakes

import matplotlib.pyplot as plt
import numpy as np

months = ['Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun']
sales  = [98500, 99200, 98800, 99500, 99700, 100200]

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 5))

# WRONG: Truncated Y-axis makes tiny difference look huge
ax1.bar(months, sales, color='#F44336')
ax1.set_ylim(98000, 100500)  # 2.2% range looks like 200% growth!
ax1.set_title('❌ Truncated Y-Axis\n(2.2% difference looks massive)', color='red')

# RIGHT: Full Y-axis shows true magnitude
ax2.bar(months, sales, color='#4CAF50')
ax2.set_ylim(0, 120000)
ax2.set_title('✅ Full Y-Axis\n(True magnitude preserved)', color='green')
ax2.text(0.5, 0.95, 'Actual growth: +1.7%', transform=ax2.transAxes,
          ha='center', fontsize=11, color='green', fontweight='bold')

plt.suptitle('Mistake: Truncated Y-Axis', fontsize=13, fontweight='bold')
plt.tight_layout()
plt.savefig('truncated_axis.png', dpi=150)
plt.show()

# Using a line chart for unrelated categories (not time series)
categories = ['Marketing', 'Engineering', 'Sales', 'HR', 'Finance']
headcount  = [25, 48, 32, 12, 18]

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 5))

# WRONG: Line chart implies continuous change between categories
ax1.plot(categories, headcount, 'o-', color='red', linewidth=2, markersize=8)
ax1.set_title('❌ Wrong: Line Chart for Categorical Data\n(Line implies connection between depts)', color='red')

# RIGHT: Bar chart for categorical comparison
ax2.bar(categories, headcount, color=['#1565C0' if h == max(headcount) else '#90CAF9' for h in headcount])
for i, val in enumerate(headcount):
    ax2.text(i, val + 0.3, str(val), ha='center', fontweight='bold')
ax2.set_title('✅ Correct: Bar Chart for Categories', color='green')
ax2.spines['top'].set_visible(False)
ax2.spines['right'].set_visible(False)

plt.suptitle('Mistake: Wrong Chart Type', fontsize=13, fontweight='bold')
plt.tight_layout()
plt.savefig('wrong_chart_type.png', dpi=150)
plt.show()

# Too many slices / no sorting
labels = [f'Cat {i}' for i in range(1, 13)]  # 12 categories!
sizes  = np.random.dirichlet(np.ones(12)) * 100

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

# WRONG: 12-slice pie
ax1.pie(sizes, labels=labels, autopct='%1.1f%%', startangle=90)
ax1.set_title('❌ Wrong: 12-Slice Pie\n(Impossible to compare)', color='red')

# RIGHT: Sorted bar chart
sorted_pairs = sorted(zip(sizes, labels), reverse=True)
s_sorted, l_sorted = zip(*sorted_pairs)
ax2.barh(l_sorted[::-1], list(s_sorted)[::-1],
          color=['#1565C0' if s == max(s_sorted) else '#90CAF9' for s in reversed(list(s_sorted))])
ax2.set_xlabel('Share (%)')
ax2.set_title('✅ Better: Sorted Horizontal Bar', color='green')
ax2.spines['top'].set_visible(False)
ax2.spines['right'].set_visible(False)

plt.suptitle('Mistake: Too Many Pie Slices', fontsize=13, fontweight='bold')
plt.tight_layout()
plt.savefig('pie_mistake.png', dpi=150)
plt.show()

MISTAKE                       FIX
──────────────────────────────────────────────────────────────────
1. Truncated Y-axis           → Start at 0 for bar charts
2. Wrong chart type           → Use chart selection decision tree
3. Too many pie slices (>5)   → Use sorted bar chart or treemap
4. 3D charts                  → Never use 3D for 2D data
5. Rainbow (jet) colormap     → Use viridis or perceptually uniform
6. Too many colors            → Max 7 categories; grey for context
7. Missing labels/units       → Always label axes with units
8. Dual Y-axes                → Use separate charts instead
9. Unsorted bars              → Sort by value (except time data)
10. Overplotting              → Use alpha, hexbin, or sampling