Assignment 1: Descriptive analytics report¶

Your first written analytics report. You run a given Python script on a restaurant's transaction records (bills, tips, party sizes, days and shifts), which prints two summary tables and draws three figures, and you turn what they show into a four-section business report for the restaurant's owner. You submit the report and the executed notebook as one PDF on GAP.

Adapted from the DATA 300 GAP course, In-Class Module 1, Assignment 1: Data Visualization and Descriptive Analytics Report. Written in our own words; follow GAP for the official wording.

1. What and why¶

The data has one row per restaurant bill (a transaction): total_bill and tip in US dollars, the server's sex, whether the party included a smoker, the day (Thur, Fri, Sat, Sun), the time (Lunch or Dinner) and the party size (number of diners).

2. Route A: Try JupyterLab (as on GAP)¶

Part 1: Open and name a notebook¶

Open Try JupyterLab, start a Python (Pyodide) notebook and rename it YourLastName_DATA300_Assignment1.ipynb.

Part 2: Run the script¶

Paste GAP's whole script into the first cell and press Shift + Enter. In outline, it:

1. installs seaborn, pandas and matplotlib with %pip, and loads tips.csv from GitHub with open_url;
2. prints Table 1: the mean bill, mean tip and mean party size for each day;
3. prints Table 2: the total of all bills for each shift (Lunch and Dinner);
4. draws Figure 1: a bar chart of total bills by day;
5. draws Figure 2: a box plot of tips for each shift;
6. draws Figure 3: a scatter plot of tip against bill, with a fitted (OLS) regression line and a shaded band.

Check that both tables and all three figures appear before you start writing.

Part 3: Write the report¶

Write the report in Word (or Markdown) using GAP's four headings exactly; see How to write the report below. Use paragraphs, except for the two recommendations, which are bullet points.

Part 4: Export¶

Run the notebook from the top, then File → Print → Save as PDF with Background graphics ticked, named YourLastName_Jupyter.pdf. Save the report as YourLastName_Report.pdf. See Submission below.

How to write the report¶

The four headings, exactly as GAP gives them, in this order: 1. Executive Summary, 2. Descriptive Data Overview, 3. Visual Insights & Graphical Trends, 4. Strategic Business Recommendations. What goes under each:

1. Executive Summary: one paragraph. The client's objective, who you are (Bistro Analytics), the data analysed (the tips dataset and its number of transactions; check the count with len(df)), and the two operational goals: scheduling staff for peak demand, and growing revenue while keeping good servers. Lead with your main finding and what you recommend, in one or two sentences (the finding-first rule from the morning lecture).
2. Descriptive Data Overview: one or two paragraphs. From Table 1, which day has the highest average bill and which the highest average tip, with the values. From Table 2, total revenue for Lunch and for Dinner and how far apart they are.
3. Visual Insights & Graphical Trends: two or three paragraphs, one per figure. Figure 1: compare the weekday days in the data with the weekend days and name the peak days. Figure 2: compare the medians and interquartile ranges of tips at Lunch and Dinner, discuss what that means for a server's earnings, and comment on the points beyond the upper whiskers. Figure 3: describe the slope and direction of the regression line and what it suggests for upselling.
4. Strategic Business Recommendations: two detailed bullet points. Recommendation 1 (staffing): which days and shifts should get the strongest servers, and why. Recommendation 2 (upselling and training): a concrete programme justified by the bill-to-tip relationship, explaining how a higher bill benefits both the restaurant and the server.

Evidence and numbers. Every claim should point to a table or figure and carry a number with its unit:

The brackets, X and Y are placeholders: fill them from your own output, and only with what it shows. Write money as US dollars ($ or "US dollars"), say "mean" or "median" rather than "average" when it matters, and round sensibly (to the cent for means, to the dollar for totals).

Reading the box plot (Figure 2). The line inside each box is the median tip; the box covers the middle half of the tips (the interquartile range, from the 25th to the 75th percentile); the whiskers reach the furthest tips within 1.5 × IQR of the box; dots beyond them are outliers: unusually large tips, not errors. Compare the medians (typical tip), the box heights (how predictable tips are) and the outliers (occasional large tips), and say what each means for a server choosing shifts. The box plot shows tip amounts per bill, not total tips per shift: a shift with more bills can earn a server more in total even with similar typical tips.

Reading the regression line (Figure 3). The slope is the average change in tip (US dollars) for each one-dollar increase in the bill. To read it from the chart, take two points on the line far apart, divide the rise in tip by the rise in bill, and state it in words ("about $X more tip for each extra $10 of bill"). Then: say over what range of bills it holds (do not extrapolate far beyond the data); say how widely the points scatter around the line; and remember it is an association: larger bills go with larger tips, but larger parties, longer meals and the day also differ between bills. The shaded band is the uncertainty of the line, not a range for individual tips. Route B below can print the slope for you.

Recommendations that are actionable. Each one says who does what, when, why (citing your evidence) and how you would know it worked (a metric to watch, such as average bill per table or tip per server shift). Keep them within what the data shows; for example, the data does not include staff costs or how many servers worked each shift, so say so if it limits the advice.

3. Route B: offline in Anaconda JupyterLab¶

Use this route if Try JupyterLab cannot load the data, or if you prefer your own laptop. It produces the same two tables and three figures from the bundled copy of the data, with nothing installed or downloaded. Create a notebook inside the DATA300_GAP folder (next to data), named as in Route A, and paste the block into one cell (or split it at the # --- lines).

# Route B for Assignment 1: run offline in Anaconda JupyterLab.
# The tips file comes from the course folder, so nothing is installed or downloaded.
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

df = pd.read_csv("data/tips.csv")

# Put days and shifts in calendar order (GAP's version lists the days alphabetically;
# the numbers are the same, only the order differs)
df["day"] = pd.Categorical(df["day"], categories=["Thur", "Fri", "Sat", "Sun"], ordered=True)
df["time"] = pd.Categorical(df["time"], categories=["Lunch", "Dinner"], ordered=True)
print("Transactions in the data:", len(df))

# --- Table 1: average transaction metrics by day of week
day_summary = df.groupby("day", observed=True)[["total_bill", "tip", "size"]].mean().round(2)
day_summary.columns = ["Avg Total Bill ($)", "Avg Tip ($)", "Avg Party Size"]
print("\n--- Table 1: Average Transaction Metrics by Day ---")
print(day_summary)

# --- Table 2: total revenue by meal shift
time_summary = df.groupby("time", observed=True)[["total_bill"]].sum().round(2)
time_summary.columns = ["Total Revenue ($)"]
print("\n--- Table 2: Aggregate Revenue Breakdown by Meal Shift ---")
print(time_summary)

sns.set_theme(style="whitegrid")

# --- Figure 1: total revenue by day (one colour, works on any seaborn version)
rev_by_day = df.groupby("day", observed=True)["total_bill"].sum().reset_index()
plt.figure(figsize=(8, 5))
sns.barplot(data=rev_by_day, x="day", y="total_bill", color="#66c2a5")
plt.title("Figure 1: Total Gross Revenue Generated by Day of Week", fontsize=14, fontweight="bold")
plt.xlabel("Day of Week")
plt.ylabel("Total Revenue ($)")
plt.tight_layout()
plt.show()

# --- Figure 2: tip distribution for each shift
plt.figure(figsize=(8, 5))
sns.boxplot(data=df, x="time", y="tip", color="#a1c9f4")
plt.title("Figure 2: Distribution of Server Tip Amounts: Lunch vs. Dinner", fontsize=14, fontweight="bold")
plt.xlabel("Meal Shift")
plt.ylabel("Tip Amount ($)")
plt.tight_layout()
plt.show()

# --- Figure 3: tip against bill with a fitted regression line
plt.figure(figsize=(9, 6))
sns.regplot(data=df, x="total_bill", y="tip", seed=0,
            scatter_kws={"alpha": 0.6}, line_kws={"color": "red", "linewidth": 2})
plt.title("Figure 3: Linear Relationship: Total Bill Size vs. Tip Amount", fontsize=14, fontweight="bold")
plt.xlabel("Total Bill Amount ($)")
plt.ylabel("Tip Amount ($)")
plt.tight_layout()
plt.show()

# --- Optional: the numbers behind Figure 3's line, for your report
slope, intercept = np.polyfit(df["total_bill"], df["tip"], 1)
r = df["total_bill"].corr(df["tip"])
print(f"Fitted line: tip = {intercept:.2f} + {slope:.3f} x total_bill   (US dollars)")
print(f"Correlation between bill and tip: r = {r:.2f}")
print(f"Bills in the data range from ${df['total_bill'].min():.2f} to ${df['total_bill'].max():.2f}")

In JupyterLab you may replace print(day_summary) and print(time_summary) with display(...) for nicer tables. Export with Kernel → Restart Kernel and Run All Cells, then File → Print → Save as PDF.

4. Checks you can make¶

These tell you whether your output and report are complete without giving any answers away.

• Table 1 has four rows (Thur, Fri, Sat, Sun) and three columns; money columns are in US dollars, party size in diners.
• Table 2 has two rows (Lunch and Dinner); the two totals add up to the sum of all bills, df["total_bill"].sum().
• Figure 1 has four bars; Figure 2 has two boxes; Figure 3 has one rising or falling line through a cloud of points, with a shaded band.
• The report has the four headings exactly as GAP words them, in order, with paragraphs (and bullets only for the recommendations).
• Every number in the report can be found in, or read from, a table or figure, and carries a unit.
• Section 3 mentions medians, the IQR and outliers for Figure 2, and the slope in "dollars of tip per dollar (or per $10) of bill" for Figure 3.
• Each recommendation names specific days and shifts or a specific programme, cites evidence, and says how success would be measured.
• The PDF has the report first, then the executed notebook.

5. Known pitfalls and fixes¶

6. Beyond the assignment¶

Optional, not graded. Try these after you have written your report:

1. Count the bills for each day and shift with pd.crosstab(df["day"], df["time"]). Which day–shift combinations exist in the data, and what does that mean for comparing Lunch with Dinner?
2. Compute the tip rate two ways: df["tip"].sum() / df["total_bill"].sum() (ratio of sums) and (df["tip"] / df["total_bill"]).mean() (mean of ratios). Why do they differ (morning lecture, Part 6)?
3. Draw Figure 3 again with hue="size" or with col="time" in sns.lmplot. Does the bill-to-tip relationship look the same for small and large parties?
4. Make Figure 1 show the mean bill per transaction instead of the total. Which chart better supports a staffing decision, and which an upselling decision?
5. Rewrite each figure's title as a message title that states your finding.

7. Submission¶

1. Restart and run the notebook from the top; check that both tables and all three figures appear.
2. Export it with File → Print → Save as PDF, Background graphics ticked (or File → Save and Export Notebook As → HTML, then print the HTML to PDF).
3. Combine the report and the notebook PDF offline, report first (Combining PDFs without internet).
4. Open the final PDF and check both parts are complete and readable.
5. Upload it to Assignment 1 on GAP and submit before Sunday 15 November, 23:59 China Standard Time (how to submit).