Questions from Last Week?

• Manipulating & cleaning DataFrames?
• Aggregating data with groupby?
• Joining datasets to bring in context?
• Anything from the quiz or lab?

Why We Visualize

• Communicate insights quickly — a well-chosen chart reveals in seconds what a table hides in rows
• Reveal patterns and outliers — shape, spread, and gaps are invisible in raw numbers
• Support decision-making — stakeholders act on visuals, not on summary statistics alone
• Expose data quality issues — unexpected spikes, missing ranges, and impossible values show up immediately in a plot

How Quickly Can You Spot Insights?

What can you tell from this table?

How Quickly Can You Spot Insights?

What can you tell from this plot?

Different Visualization Needs

Match the Tool to the Goal

We should match how we visualize to the goal we have in mind.

Pandas for Quick EDA

• First reach when you need a fast answer — minimal code, tied directly to DataFrames
• Enough for exploratory work; not meant for final deliverables
• When to use: speed over polish

# One-liner from aggregated data
basket_spend['basket_spend'].plot(kind='hist', bins=40, title='Basket Spend Distribution')

The .plot() Mental Model

One variable — call .plot() on a Series, specify kind:

series.plot(kind='hist')        # distribution of one column
series.plot(kind='box')         # spread and outliers
series.plot(kind='line')        # trend (index is x-axis)

df.plot(kind='scatter', x='col_a',    y='col_b')   # relationship
df.plot(kind='bar',     x='category', y='value') # comparison
df.plot(kind='line',    x='date',     y='value') # trend over time
df.plot(kind='barh',    x='category', y='value') # horizontal bars (long labels)

Chart Types: Distribution

basket_spend['basket_spend'].plot(
    kind='hist', bins=40, figsize=(10, 3.2),
    title='Distribution of basket spend', xlabel='Basket spend ($)'
)
plt.tight_layout()
plt.show()

Chart Types: Category Comparison

category_totals.plot(
    kind='barh', x='department', y='sales_value',
    figsize=(10, 3.5), legend=False,
    title='Total revenue by department (top 10)', xlabel='Total sales ($)'
)
plt.tight_layout()
plt.show()

Chart Types: Trend Over Time

weekly_sales.plot(
    kind='line', x='week', y='total_sales',
    figsize=(10, 3.2), legend=False,
    title='Weekly total sales', ylabel='Total sales ($)'
)
plt.tight_layout()
plt.show()

Pandas is Just a Wrapper

Pandas .plot() calls Matplotlib under the hood — which means you can mix them:

plt.style.use('fivethirtyeight')

weekly_sales.plot(
    kind='line', x='week', y='total_sales',
    figsize=(10, 3), legend=False, title='Weekly total sales'
)
plt.tight_layout()
plt.show()

Seaborn for Statistical Visualization

• Built on Matplotlib — cleaner syntax for common statistical chart types
• Shines when you need: grouping, ordering, distributions with shape
• One function call handles what Matplotlib needs 10–15 lines to produce

The Seaborn Mental Model

Every Seaborn function follows the same pattern:

sns.function(data=df, x='col', y='col', hue='group', ...)

sns.histplot(data=df, x='col', kde=True)                     # distribution shape
sns.boxplot(data=df, x='group', y='col', order=[…])          # spread across groups
sns.scatterplot(data=df, x='col_a', y='col_b', hue='group')  # relationship
sns.heatmap(pivot_df, cmap='YlOrRd')                         # matrix of values
sns.barplot(data=df, x='group', y='col', order=[…])          # mean + confidence interval

Distribution: histplot

fig, ax = plt.subplots(figsize=(10, 3.2))
sns.histplot(
    data=basket_demo,
    x='basket_spend',
    bins=50,
    kde=True,          # overlay a density curve
    ax=ax
)
ax.set_xlabel('Basket spend ($)')
ax.set_title('Distribution of basket spend — with density curve')
ax.set_xlim(0, 100)
plt.tight_layout()
plt.show()

Group Comparison: boxplot

fig, ax = plt.subplots(figsize=(10, 3.8))
sns.boxplot(
    data=basket_demo,
    x='income',
    y='basket_spend',
    order=income_order,   # explicit ordering — critical for income brackets
    ax=ax,
    showfliers=False
)
ax.set_xticklabels(ax.get_xticklabels(), rotation=45, ha='right', fontsize=8)
ax.set_xlabel('Income bracket')
ax.set_ylabel('Basket spend ($)')
ax.set_title('Basket spend by income bracket')
plt.tight_layout()
plt.show()

Patterns: heatmap

fig, ax = plt.subplots(figsize=(10, 3.8))
sns.heatmap(
    heatmap_data.iloc[:, 6:22],   # hours 6am–10pm
    cmap='YlOrRd',
    linewidths=0.3,
    ax=ax,
    cbar_kws={'label': 'Trip count'}
)
ax.set_xlabel('Hour of day')
ax.set_ylabel('')
ax.set_title('Shopping trips by day of week and hour')
plt.tight_layout()
plt.show()

When to Reach for Seaborn

Use Seaborn when you want to:

• Compare distributions across groups — histplot, boxplot, violinplot
• Visualize relationships — scatterplot with hue, lmplot
• Show patterns in a matrix — heatmap
• Do any of the above with ordered categorical variables

Matplotlib for Full Control

• The most widely used Python plotting library — and the foundation for Seaborn and Pandas
• Full customization: labels, annotations, tick formatting, styles, multi-panel layouts, export
• When to use: precision and polish for reports and presentations

The Figure / Axes Hierarchy

Everything in Matplotlib is an object you can reference and modify:

fig, ax = plt.subplots(figsize=(10, 2.5))
# fig = the overall canvas
# ax  = the actual plot area (axes, ticks, lines, labels)
fig.set_facecolor('#eaf4fb')
ax.set_facecolor('#d0eaf8')
ax.set_title('fig (blue border) contains ax (darker blue plot area)')
plt.tight_layout()
plt.show()

Step 0: Quick Pandas Starting Point

Start with a rough Pandas plot — fast, but no formatting:

weekly_sales.plot(kind='line', x='week', y='total_sales', figsize=(10, 3.0), legend=False)
plt.show()

Step 1: Move to the Figure / Axes API

Get handles — now you control everything:

fig, ax = plt.subplots(figsize=(10, 3.0))
ax.plot(weekly_sales['week'], weekly_sales['total_sales'], linewidth=2)
ax.set_title('Weekly total sales')
ax.set_xlabel('Week')
ax.set_ylabel('Total sales ($)')
plt.tight_layout()
plt.show()

Step 2: Executive-Ready Formatting

fig, ax = plt.subplots(figsize=(10, 3.2))
ax.plot(weekly_sales['week'], weekly_sales['total_sales'], linewidth=2)
ax.set_title('Weekly total sales — Regork grocery chain', pad=10)
ax.set_xlabel('Week')
ax.yaxis.set_major_formatter(mtick.StrMethodFormatter('${x:,.0f}'))
ax.grid(True, alpha=0.3)
for spine in ['top', 'right']:
    ax.spines[spine].set_visible(False)
plt.tight_layout()
plt.show()

Step 3: Highlight the Insight

fig, ax = plt.subplots(figsize=(10, 3.5))
ax.plot(weekly_sales['week'], weekly_sales['total_sales'], linewidth=2)
ax.set_title('Weekly total sales — note holiday spike in late December', pad=10)
ax.set_xlabel('Week')
ax.yaxis.set_major_formatter(mtick.StrMethodFormatter('${x:,.0f}'))
ax.grid(True, alpha=0.3)
for spine in ['top', 'right']:
    ax.spines[spine].set_visible(False)

peak_idx = weekly_sales['total_sales'].idxmax()
peak_week = weekly_sales.loc[peak_idx, 'week']
peak_val  = weekly_sales.loc[peak_idx, 'total_sales']

ax.annotate(
    f'Holiday spike\n${peak_val:,.0f}',
    xy=(peak_week, peak_val),
    xytext=(peak_week - pd.Timedelta(weeks=8), peak_val * 0.97),
    arrowprops=dict(arrowstyle='->', lw=1.2),
    fontsize=9
)
plt.tight_layout()
plt.show()

Matplotlib Is the Foundation

Most Python visualization libraries are built on top of Matplotlib — so once you know the Figure/Axes API, you can refine plots from Seaborn, Pandas, and others using the exact same code.

# --- Seaborn draws the plot ---
fig, ax = plt.subplots(figsize=(10, 3.8))
sns.boxplot(
    data=basket_demo, x='income', y='basket_spend',
    order=income_order, ax=ax, showfliers=False,
    color='steelblue'
)

# --- Matplotlib polishes it ---
ax.set_xticklabels(ax.get_xticklabels(), rotation=45, ha='right', fontsize=8)
ax.set_xlabel('Income bracket')
ax.set_ylabel('Basket spend ($)')
ax.set_title('Basket spend by income — same Seaborn call, Matplotlib finishing touches')
ax.yaxis.set_major_formatter(mtick.StrMethodFormatter('${x:,.0f}'))
for spine in ['top', 'right']:
    ax.spines[spine].set_visible(False)
plt.tight_layout()
plt.show()

Common Gotchas

Bokeh for Interactive Visualization

• Renders HTML/JavaScript — charts live in a browser, not just a notebook
• Users can zoom, pan, hover, and filter without writing new code
• Best for: dashboards, stakeholder exploration tools, final deliverables that need to be shared

Which Tool When?

EDA Is a Conversation With Your Data

Exploratory Data Analysis is not a checklist of charts to run.

It is a sequence of questions — each answer raising the next question.

A Framework for EDA

flowchart LR
    A["1️⃣ Question"] --> B["2️⃣ Structure"]
    B --> C["3️⃣ Distributions"]
    C --> D["4️⃣ Segmentation"]
    D --> E["5️⃣ Story"]
    style A fill:#2c3e50,color:#fff
    style B fill:#2980b9,color:#fff
    style C fill:#27ae60,color:#fff
    style D fill:#e67e22,color:#fff
    style E fill:#8e44ad,color:#fff

Step 1: Start With a Sharp Question

The question determines everything that follows.

Steps 2–3: Structure → Distributions

Check your data before trusting it:

# Step 2: understand the shape
print(f"Baskets: {basket_spend['basket_id'].nunique():,}")
print(f"Households with demographics: {basket_demo['household_id'].nunique():,}")
print(f"Median basket spend: ${basket_demo['basket_spend'].median():.2f}")

Baskets: 155,848
Households with demographics: 801
Median basket spend: $19.91

Step 4: Segmentation — Where the Story Lives

Distributions tell you what. Segmentation tells you who and where.

Step 5: Build the Story

A good EDA concludes with something worth saying:

• What did you find that you did not expect?
• What would you recommend based on this finding?
• What would you investigate next?

Before Thursday: Read Chapter 15

Thursday’s lab is the project ideation session — you’ll brainstorm and pitch.

To get the most out of it, read Chapter 15 (EDA) before lab.

It walks through a full EDA case study on the Complete Journey data and shows the framework in action from first question to final story. Coming in with that mental model will make Thursday’s brainstorming sharper.

The Regork Scenario

You are a data scientist at Regork, a national grocery chain.

Your manager has asked you to identify a potential growth opportunity — an area where the company could invest future resources to increase revenue or profits.

What a Sharp Business Question Looks Like

What Strong Analysis Looks Like

Clear business problem — state one specific, answerable question tied to a real decision. Not “understand customer behavior.” Something like: “Do low-income households redeem coupons at higher rates, and does redemption translate into larger basket spend or just a cheaper version of the same basket?”

Proposed value to the decision-maker — name who acts on your findings and what they would do differently. The VP of Promotions needs a recommendation, not a summary of what you computed.

Logical, methodical analysis — start with data structure and distributions, then move into group comparisons and segmentation. Each step should raise the next question. Show your reasoning, not just your outputs.

Clearly explained findings — every chart and table earns its place by making one point. Label axes properly, use readable scales, and narrate what the reader should see. Plots without interpretation are just decoration.

Polished communication — the report reads as a coherent story, not a list of code cells. The presentation looks like a business deck, not a notebook printout. Your audience is an executive, not a grader.

Clear recommendations — end with something actionable. “Based on our findings, we recommend…” is a sentence your report and presentation must contain.

Deliverables

A. Written Technical Report

• Jupyter Notebook rendered to HTML
• Must include: data preparation, business question, EDA, narrative findings, recommendations
• Show relevant code; suppress warnings and distracting output
• File naming: YYYY_BANA7025_groupXX_project.html

B. Recorded Stakeholder Presentation

• 3 minutes maximum — no code, executive audience
• Covers: business problem, key findings, one clear recommendation
• Recorded via Zoom or Kaltura and submitted through Canvas

Before You Start — Read the Project Details

Thursday’s Lab: Project Ideation

In Thursday’s lab you will:

1. Form your group (2–4 students) and join a Canvas group
2. Brainstorm three distinct project ideas, each with a business problem, decision-maker, data plan, and success criteria
3. Choose your most compelling idea and do a 2–3 minute stand-up pitch to the class

Key Takeaways

Part 1 — Python Visualization

• Match the tool to the goal: Pandas for speed, Seaborn for statistical grouping, Matplotlib for polish, Bokeh for interactivity
• Pandas is a wrapper over Matplotlib — knowing both gives you full control