Create Visualization Skill

Create publication-quality data visualizations using Python. Select the right chart type, apply design principles, and produce clean, accessible figures ready for reports and presentations.

Workflow

Step 1: Understand the Data and Message

Before choosing a chart type, answer:

1. What is the message? Every chart should make one point clearly.
2. What type of data? Categorical, continuous, temporal, geospatial?
3. What relationship? Trend, comparison, composition, distribution, correlation?
4. How many variables? One, two, three or more dimensions?
5. How many data points? 5 categories vs. 5,000 observations require different approaches.
6. Where will it be displayed? Slide deck, printed report, web page, notebook?

Step 2: Select the Chart Type

Charts to avoid:

• Pie charts with more than 5 slices
• 3D charts (distort perception)
• Dual y-axis charts (confuse readers) -- use two separate charts instead
• Radar/spider charts (hard to read)

Step 3: Prepare the Data

import pandas as pd
import numpy as np

# Standard data preparation
df = pd.read_csv('data.csv', parse_dates=['date'])
df = df.sort_values('date')

# Aggregation for visualization
summary = df.groupby('category').agg(
    total=('revenue', 'sum'),
    count=('order_id', 'nunique')
).reset_index().sort_values('total', ascending=False)

Step 4: Build the Visualization

Matplotlib/Seaborn (Static, Publication-Quality)

import matplotlib.pyplot as plt
import matplotlib.ticker as mticker
import seaborn as sns

# Set global style
plt.style.use('seaborn-v0_8-whitegrid')
sns.set_palette('colorblind')

fig, ax = plt.subplots(figsize=(10, 6))

# Plot
ax.plot(df['date'], df['revenue'], linewidth=2, color='#2563eb')

# Formatting
ax.set_title('Monthly Revenue Trend', fontsize=16, fontweight='bold', pad=16)
ax.set_xlabel('')
ax.set_ylabel('Revenue ($)', fontsize=12)
ax.yaxis.set_major_formatter(mticker.FuncFormatter(lambda x, _: f'${x/1e6:.1f}M'))
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)

# Annotations
ax.annotate('Product Launch', xy=(launch_date, launch_value),
            xytext=(10, 20), textcoords='offset points',
            arrowprops=dict(arrowstyle='->', color='gray'),
            fontsize=10, color='gray')

plt.tight_layout()
plt.savefig('revenue_trend.png', dpi=150, bbox_inches='tight')
plt.show()

Plotly (Interactive)

import plotly.express as px
import plotly.graph_objects as go

fig = px.line(df, x='date', y='revenue', color='segment',
              title='Revenue by Segment',
              labels={'revenue': 'Revenue ($)', 'date': ''},
              template='plotly_white')

fig.update_layout(
    font=dict(family='Inter, sans-serif', size=12),
    legend=dict(orientation='h', yanchor='bottom', y=1.02),
    hovermode='x unified'
)

fig.update_yaxes(tickprefix='$', tickformat=',.0f')
fig.write_html('revenue_interactive.html')
fig.show()

Step 5: Apply Design Principles

Color:

• Use a colorblind-friendly palette (e.g., colorblind, Set2, or custom)
• Limit to 5-7 colors maximum
• Use color intensity to encode magnitude
• Reserve red/green for clear positive/negative semantics
• Use gray for context/baseline, color for the focus

Typography:

• Title: 14-18pt, bold
• Axis labels: 11-13pt
• Tick labels: 9-11pt
• Annotations: 9-11pt, lighter color
• Use sentence case, not ALL CAPS

Layout:

• Remove chart junk: unnecessary gridlines, borders, backgrounds
• Maximize data-ink ratio
• Left-align text elements
• Add whitespace between elements
• Use tight_layout() or constrained_layout=True

Accessibility:

• Do not rely on color alone -- use patterns, markers, or labels
• Ensure minimum contrast ratio of 4.5:1 for text
• Add alt text when embedding in HTML or documents
• Use at least 2pt line width for line charts

Step 6: Save and Export

# High-resolution PNG for reports
fig.savefig('chart.png', dpi=300, bbox_inches='tight', facecolor='white')

# SVG for presentations (scalable)
fig.savefig('chart.svg', bbox_inches='tight')

# PDF for print
fig.savefig('chart.pdf', bbox_inches='tight')

# Interactive HTML (plotly)
fig.write_html('chart.html', include_plotlyjs='cdn')

Output Format

Deliver:

1. The Python code that generates the visualization (reproducible)
2. The rendered image or HTML file
3. A brief note on chart type selection rationale
4. Suggestions for alternative views if relevant

Examples

Example 1: Time Series Trend

Line chart with confidence bands, annotated events, formatted axis labels, reference line for target.

Example 2: Category Comparison

Horizontal bar chart sorted by value, value labels on bars, highlighted top-N, muted colors for the rest.

Example 3: Distribution Analysis

Histogram with KDE overlay, vertical lines for mean/median, quartile annotations, sample size noted.

Example 4: Correlation

Scatter plot with regression line, R-squared annotation, marginal histograms, color-coded by segment.

Best Practices

1. One message per chart. If you need to show two things, make two charts.
2. Title states the insight, not the data. "Revenue grew 23% in Q4" is better than "Q4 Revenue."
3. Sort meaningfully. Sort bars by value, not alphabetically. Sort time series chronologically.
4. Start y-axis at zero for bar charts. Line charts may use a non-zero baseline if justified.
5. Format numbers for humans. Use $1.2M not 1200000. Use thousands separators.
6. Remove legends when possible. Label data directly on the chart instead.
7. Use consistent scales when comparing. Side-by-side charts should share the same axis range.
8. Test in grayscale. Print a black-and-white version to ensure the chart still works.
9. Size charts for the medium. Slides need larger text than notebooks. Print needs higher DPI than screens.
10. Include data source. Add a small footnote with the source and date range.