Topic 7.6: The Pipeline Payoff

In Week 6 you learned NumPy — fast numerical computation on arrays — and Pandas — labeled data structures for analysis. In Week 7 you have been learning Matplotlib — turning numbers into charts. This topic is where all three converge.

The critical insight is that you do not need to manually convert between these libraries. Matplotlib understands Pandas Series and NumPy arrays natively. You can pass a DataFrame column directly into ax.bar() or ax.plot() — no loops, no list conversion, no intermediate steps.

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

# Load the Egyptian cities transport dataset
transport = pd.read_csv('../Datasets/7_6_transport_stats.csv')

print("Transport Stats Dataset:")
print(transport)
print("\nShape:", transport.shape)
print("Columns:", list(transport.columns))

Before Pandas, creating a bar chart from data required loading values into a list, iterating through rows, and managing index positions manually. With Pandas, a single column is all Matplotlib needs — no loop, no conversion.

When you pass transport['city'] as the x-axis argument, Matplotlib interprets the Pandas Series as a sequence of category labels. When you pass transport['daily_passengers_m'] as the y-axis argument, it reads the numeric values. The library handles everything automatically.

# Plot: daily passengers per city (bar chart using DataFrame column directly)
fig, ax = plt.subplots(figsize=(12, 5))

ax.bar(transport['city'], transport['daily_passengers_m'],
       color='steelblue', alpha=0.85)

ax.set_title('Daily Passengers by Egyptian City (Millions)', fontsize=14, fontweight='bold')
ax.set_xlabel('City')
ax.set_ylabel('Daily Passengers (Millions)')
ax.tick_params(axis='x', rotation=45)

plt.tight_layout()
plt.show()

print("DataFrame column fed directly into ax.bar() — no loop, no manual list")

Pandas Series and NumPy arrays are interchangeable in most Matplotlib operations. But when you explicitly need a NumPy array — perhaps for mathematical operations before plotting — the .to_numpy() method converts any Pandas Series to a bare NumPy ndarray.

# Extract columns as NumPy arrays
speed     = transport['avg_speed_kmh'].to_numpy()
emissions = transport['co2_emissions_kt'].to_numpy()

print(f"Type of speed: {type(speed)}")

# Scatter: speed vs emissions
fig, ax = plt.subplots(figsize=(10, 6))

ax.scatter(speed, emissions, alpha=0.7)

ax.set_title('Speed vs. CO₂ Emissions — Egyptian Cities', fontsize=14, fontweight='bold')
ax.set_xlabel('Average Speed (km/h)')
ax.set_ylabel('CO₂ Emissions (kt)')

plt.tight_layout()
plt.show()

Both transport['avg_speed_kmh'] (a Pandas Series) and speed (a NumPy array) would work identically as input to ax.scatter(). The explicit conversion with .to_numpy() is useful when you need to apply NumPy functions to the data before plotting — like np.log(), np.cumsum(), or array slicing with pure NumPy syntax.

Matplotlib's default appearance is functional but plain. The library ships with dozens of built-in themes that apply a complete visual makeover to any chart — changing background, gridlines, color palettes, and typography — in a single line of code.

# Show available styles
print("Available styles (first 10):")
print(plt.style.available[:10])
print()

# Apply a style and redraw the bar chart
plt.style.use('ggplot')

fig, ax = plt.subplots(figsize=(12, 5))

ax.bar(transport['city'], transport['daily_passengers_m'],
       color='steelblue', alpha=0.9)

ax.set_title('Daily Passengers — ggplot style', fontsize=13, fontweight='bold')
ax.set_xlabel('City')
ax.set_ylabel('Daily Passengers (Millions)')
ax.tick_params(axis='x', rotation=45)

plt.tight_layout()
plt.show()

# Reset to default
plt.style.use('default')
print("Style applied and reset to default.")

plt.show() displays a chart inside a Jupyter Notebook. But the notebook is not a deliverable — a manager, client, or colleague who doesn't have Python cannot open it. fig.savefig() writes the chart as a real image file (PNG, PDF, SVG) that can be embedded in any document or presentation.

# Reuse the bar chart figure from above and save it to disk
fig.savefig('transport_report.png', dpi=150, bbox_inches='tight')

plt.show()
plt.style.use('default')

The two most important parameters are dpi and bbox_inches. DPI (dots per inch) controls resolution: 150 dpi is good for screen display; 300 dpi is standard for print. bbox_inches='tight' trims whitespace from the edges and ensures no labels are cut off — without it, axis labels and titles near the figure border may be partially clipped.