Final Exam

Version A

Published

December 12, 2025


Section 1. Multiple Choice

Question 1

Categorical data distributions are commonly visualized using histograms, while numerical data distributions are commonly shown with bar charts.

  • True
  • False

False.
Histograms are typically used for numerical (quantitative) variables, while bar charts are typically used for categorical variables.


Question 2

The popularization of sports analytics was significantly influenced by the “Moneyball” book published in 2003 and its subsequent movie adaptation.

  • True
  • False

True.


Question 3

Why are tools like Excel or Google Sheets not considered full Database Management Systems (DBMS)?

  1. They cannot store numerical data.
  2. They provide basic storage but lack robust capabilities for querying, updating consistency, and managing large-scale data safely.
  3. They do not support the creation of charts or visualizations.
  4. They are incompatible with CSV files.

b. They provide basic storage but lack robust capabilities for querying, updating consistency, and managing large-scale data safely.


Question 4

Which of the following best describes the modern “vibe coding” workflow in data analytics?

  1. Writing all SQL and Python code manually to ensure 100% accuracy without AI intervention.
  2. Using drag-and-drop interfaces exclusively without any coding logic.
  3. Outsourcing all coding tasks to third-party distributed agents.
  4. Prompting AI assistants to generate logic flows and code snippets, then reviewing the output for accuracy.

d. Prompting AI assistants to generate logic flows and code snippets, then reviewing the output for accuracy.


Question 5

Which of the following is NOT one of the “Four Rules for Co-Intelligence” for working with AI?

  1. Always invite AI to the table
  2. Be the human in the loop (HITL)
  3. Automate everything possible and remove human oversight
  4. Treat AI like a person (but remember it isn’t)
  5. Assume this is the worst AI you’ll ever use

c. Automate everything possible and remove human oversight


Question 6

In a relational database, what is the role of a key variable?

  1. It stores raw data without any transformation.
  2. It uniquely identifies each row and helps define relationships between tables.
  3. It limits users’ access to only certain tables.
  4. It performs Map and Reduce tasks on big data sets.

b. It uniquely identifies each row and helps define relationships between tables.


Question 7

Which of the following is a characteristic of “Reinforcement Learning from Human Feedback” (RLHF)?

  1. It involves humans ranking or scoring model answers to align the model with human preferences for safety and helpfulness.
  2. It allows the model to learn entirely on its own without any human intervention.
  3. It is a pre-training phase where the model reads vast amounts of text.
  4. It is primarily used for generating images from text descriptions.

a. It involves humans ranking or scoring model answers to align the model with human preferences for safety and helpfulness.


Question 8

Which type of visualization is most suitable for showing the distribution of a single numerical variable?

  1. Bar Chart
  2. Histogram
  3. Scatterplot
  4. Line Chart

b. Histogram


Question 9

If you want to visualize the relationship between two numerical variables and see whether they move together, which chart would be most appropriate?

  1. Bar Chart
  2. Histogram
  3. Scatterplot
  4. Boxplot

c. Scatterplot


Question 10

In the ETL process, which of the following best describes the “Transform” stage’s primary purpose?

  1. Aggregating and filtering data from disparate sources before extraction.
  2. Altering, cleaning, and integrating data to ensure consistency and usability.
  3. Moving transformed data into staging environments for downstream operations.
  4. Ensuring that outdated data is removed or archived for regulatory compliance.

b. Altering, cleaning, and integrating data to ensure consistency and usability.


Question 11

What does the term “API” stand for, and what is its primary function as described in the lecture?

  1. Automated Processing Interface; it cleans messy CSV files automatically.
  2. Application Programming Interface; it allows software systems to communicate and request data programmatically.
  3. Advanced Python Integration; it translates R code into Python.
  4. Analytical Pipeline Interface; it is used exclusively for visualizing Tableau dashboards.

b. Application Programming Interface; it allows software systems to communicate and request data programmatically.


Question 12

Which statement is most accurate about tokens in the context of large language models (LLMs)?

  1. A token is always exactly one English word
  2. Tokens are always single characters
  3. Tokens exist only at the output side, not at the input side
  4. A token is a unit of text; it may be a character, a whole word, or part of a word

d. A token is a unit of text; it may be a character, a whole word, or part of a word.


Questions 13-16

For Questions 13-16, consider the following data.frame, spotify_data, displayed below:

UserID Age Gender SubscriptionTier FavoriteGenre HoursListened LastLoginTime
1 19 Male Free Pop 10.4 1.2
2 27 Female Premium Rock 15.8 22.4
3 35 Male Family Hip-Hop 22.3 13.6
4 22 Female Premium Classical 9.7 19.1
5 40 Male Free Jazz 18.6 23.5
6 31 Female Family Electronic 20.1 7.9
7 29 Male Premium Country 14.5 2.3
8 33 Female Free Blues 12.8 15.6
9 24 Female Family Reggae 17.9 20.7
10 37 Male Premium Metal 19.3 5.4
AccountMonths Satisfaction nDevices LastTrackRating nPlaylists Language
3 3 1 8.1 2 Spanish
12 5 3 7.4 4 English
18 4 2 6.8 3 French
5 2 1 9.2 1 German
20 4 3 7.5 5 English
24 5 2 8.9 3 Italian
6 3 4 6.3 4 English
10 4 1 9.0 2 English
15 5 2 8.4 3 Spanish
8 5 3 7.7 5 French

Description of Variables in netflix_data:

  1. UserID: Identifier for each user
  2. Age: Age of the user in years
  3. Gender: Gender of the user
  4. SubscriptionTier: Type of Spotify subscription
  5. FavoriteGenre: User’s favorite genre
  6. HoursListened: Average hours listened per week
  7. LastLoginTime: Time of last login in hours since midnight
  8. AccountMonths: Age of the account in months
  9. Satisfaction: User satisfaction rating (1 to 5 stars)
  10. nDevices: Number of devices connected
  11. LastTrackRating: Rating of the last played track (1.0 to 10.0)
  12. nPlaylists: Number of playlists on the account
  13. Language: User’s preferred language

Question 13

What type of variable is FavoriteGenre in the dataset?

  1. Nominal
  2. Ordinal
  3. Interval
  4. Ratio

a. Nominal


Question 14

What type of variable is SubscriptionTier in the dataset?

  1. Nominal
  2. Ordinal
  3. Interval
  4. Ratio

b. Ordinal
(There is a meaningful order: Free < Family < Premium.)


Question 15

What type of variable is LastLoginTime in the dataset?

  1. Nominal
  2. Ordinal
  3. Interval
  4. Ratio

c. Interval
(Hours since midnight: differences are meaningful, but “0” is an arbitrary reference point, not “no time.”)


Question 16

What type of variable is Satisfaction in the dataset?

  1. Nominal
  2. Ordinal
  3. Interval
  4. Ratio

b. Ordinal
(Star ratings have an order, but equal gaps between levels are not guaranteed.)


Section 2. Filling-in-the-Blanks

Question 17

In retail analytics, analyzing which products tend to sell together allows for
_______________________________ to reveal hidden product correlations and inform bundling strategies.

association rule (market basket analysis)


Question 18

A(n) ________________________________ is a visual display of key information, data, and metrics, often used in BI to provide insights at a glance.

dashboard (key performance indicator (KPI))


Question 19

________________________________ is a Python data visualization library that provides a high-level, elegant interface for creating informative and attractive graphics. You can think of it as the Python counterpart to R’s ggplot2: it emphasizes clear defaults, aesthetic color palettes, and concise syntax for complex visualizations.

Seaborn


Question 20

________________________________ is the tendency for values of two variables to vary together, and can be visualized using scatterplots.

Correlation


Question 21

________________________________ data refers to data that is not organized in a predefined manner and includes sources like social media posts, emails, photos, and videos.

Unstructured


Question 22

When designing visuals, the goal is to convey as much information as possible while minimizing _______________________________ for the audience.

cognitive load


Question 23

One of our alumni guest’s companies uses Snowflake as their _______________________________, a centralized repository that stores and manages large volumes of structured and semi-structured data for analytics and reporting purposes.

data warehouse (database management system (DBMS))


Question 24

Three most popular programming languages for data analysts are _______________________________, Python, and R.

SQL


Section 3. Data Analysis with R

Question 25

Consider the following vector x:

x <- c(2, 4, 6, 8, 10)

Write the R code to create a new vector called z, where its \(i\)-th entry (\(i = 1,2,3,4, \text{or } 5\)) is the standardized value of \(i\)-th element of x vector.

\[ z_{i} = \frac{x_{i} - \bar{x}}{\sigma_{x}} \]

  • \(\bar{x}\): the mean of values in x
  • \(\sigma_{x}\): the standard deviation of values in x

Answer: ______________________________________

z <- (x - mean(x)) / sd(x)


Question 26

Given the data.frame df with variables height and name, which of the following expressions returns a vector containing the values in the height variable?

  1. df:height
  2. df$height
  3. df::height
  4. Both b and c

b. df$height


Question 27

Consider the following data.frame, students:

Name Age Major GPA
Alice 22 Business Administration 3.8
Bob 23 Accounting 3.2
Charlie 21 Data Analytics 3.9
Diana 24 Economics 3.5

Which of the following R codes will correctly create a new data.frame with only the Name and GPA variables?

  1. students |> select(Name, GPA)
  2. students |> select(-Age, -Major)
  3. Both a and b

c. Both a and b


Question 28

Consider the following data.frame df0:

x y
Na 7
2 NA
3 9

What is the result of median(df0$y)?

  1. 7
  2. NA
  3. 8
  4. 9

b. NA


Question 29

Consider the two related data.frames, df_1 and df_2:

  • df_1
id name age
1 Bob 19
2 Julia 21
4 Zachary 20
  • df_2
id major
1 Economics
2 Business Administration
3 Data Analytics

Which of the following R code correctly join the two related data.frames, df_1 and df_2, to produce the resulting data.frame shown below?

id name age major
1 Bob 19 Economics
2 Julia 21 Business Administration
4 Zachary 20 NA
  1. df_1 |> left_join(df_2)
  2. df_2 |> left_join(df_1)
  3. Both a and b
  4. None of the above

a. df_1 |> left_join(df_2)


Questions 30-36

For Questions 30-36, consider the following R packages and the data.frame, nyc_dogs, containing individual dog license data from New York City (NYC):

library(tidyverse)
library(skimr)
library(ggthemes)

nyc_dogs <- read_csv("https://bcdanl.github.io/data/nyc_dogs_cleaned.csv")

The first 10 observations in the nyc_dogs data frame are displayed below:

name gender birth_year breed borough
paige F 2014 pit bull Manhattan
yogi M 2010 boxer Bronx
ali M 2014 basenji Manhattan
queen F 2013 akita Manhattan
lola F 2009 maltese Manhattan
ian M 2006 NA Manhattan
buddy M 2008 NA Manhattan
chewbacca F 2012 labrador Manhattan
heidi-bo F 2007 dachshund smooth coat Brooklyn
massimo M 2009 bull dog, french Brooklyn
  • The nyc_dogs data frame is with 197473 observations and 5 variables.

Description of Variables in nyc_dogs:

  • name: Dog name

  • gender: Dog gender (F for female; M for male; NA for missing value)

  • birth_year: Birth year (integer values)

  • breed: Dog breed

  • borough: Borough in NYC


The followings are the summary of the nyc_dogs data.frame, including descriptive statistics for each variable.

Data summary
Name nyc_dogs
Number of rows 197473
Number of columns 5
_______________________
Column type frequency:
character 4
numeric 1
________________________
Group variables None

Variable type: character

skim_variable n_missing min max empty n_unique
name 2637 1 30 0 26770
gender 6 1 1 0 2
breed 18832 3 35 0 295
borough 0 5 13 0 5

Variable type: numeric

skim_variable n_missing mean sd p0 p25 p50 p75 p100
birth_year 0 2012.94 4.91 1975 2010 2014 2017 2021

Question 30

What is the interquartile range of birth_year? Find this value by using the summary of the nyc_dogs data frame.

IQR = 7 = 2017 - 2010


Question 31

What R code can we use to count the number of licensed dogs by borough within each birth_year in NYC?

  1. nyc_dogs |> count(birth_year)
  2. nyc_dogs |> count(borough)
  3. nyc_dogs |> count(borough, birth_year)
  4. nyc_dogs |> count(birth_year, borough)
  5. Both c and d

e


Question 32

  • We are interested in finding the 7 most popular dog breeds in NYC.
  • To achieve this, we create a new data.frame, a new data.frame, top7_breeds, which includes only 7 most popular dog breeds, excluding any NA values.
breed n
yorkshire terrier 12804
shih tzu 12790
chihuahua 11099
labrador 11017
pit bull 10023
maltese 7750
german shepherd 5063
  • The top7_breeds data.frame is displayed above.
top7_breeds <- nyc_dogs |> 
  filter(___(1)___) |> 
  ___(2)___ |> 
  ___(3)___(-n) |> 
  head(7)  # returns the first 7 observations of the new data.frame
  • Complete the code by filling in the blanks (1)-(3).
    1. is.na(breed); (2) count(n); (3) select
    1. is.na(breed); (2) count(n); (3) arrange
    1. is.na(breed); (2) count(breed); (3) select
    1. is.na(breed); (2) count(breed); (3) arrange
    1. !is.na(breed); (2) count(n); (3) select
    1. !is.na(breed); (2) count(n); (3) arrange
    1. !is.na(breed); (2) count(breed); (3) select
    1. !is.na(breed); (2) count(breed); (3) arrange

h


Question 33

How would you describe the distribution of breed using the top7_breeds data.frame?

  • Note that the breed categories are sorted by the n variable in the plot.

Complete the code by filling in the blanks (1)-(3).

ggplot(data = top7_breeds,
       mapping = aes(___(1)___,
                     ___(2)___ = n)) +
  ___(3)___() +
  labs(y = "Top 7 Dog Breeds")

Blank (1)

  1. x = breed
  2. y = breed
  3. x = fct_reorder(n, breed)
  4. y = fct_reorder(n, breed)
  5. x = fct_reorder(breed, n)
  6. y = fct_reorder(breed, n)
  7. Both c and d
  8. Both e and f

f


Blank (2)

  1. x
  2. y
  3. fill
  4. color

a


Blank (3)

  1. geom_bar
  2. geom_col
  3. Both a and b

b


Question 34

  • We are also interested in identifying the top five most popular dog names for each gender.
  • To do this, we first create a new data frame, nyc_dogs_filtered, which includes only the observations where (1) the value of name variable is not missing and (2) the value of gender variable is not missing.
nyc_dogs_filtered <- nyc_dogs |> 
  filter(___BLANK___)
  • Which condition correctly fills in the BLANK to complete the code above?
  1. is.na(name) , is.na(gender)
  2. is.na(name) & is.na(gender)
  3. is.na(name) | is.na(gender)
  4. !is.na(name) , !is.na(gender)
  5. !is.na(name) & !is.na(gender)
  6. !is.na(name) | !is.na(gender)
  7. Both a and b
  8. Both a and c
  9. Both d and e
  10. Both d and f

i


Question 35

  • Using nyc_dogs_filtered from Question 34, we are creating the top5names_F data.frame:
name n
bella 1291
lola 1005
luna 995
lucy 914
daisy 851
  • The top5names_F data.frame provides the five most popular female dog names, displayed above.

Complete the code by filling in the blanks (1)-(2).

top5names_F <- nyc_dogs_filtered |> 
  filter(___(1)___) |> 
  count(___(2)___) |> 
  arrange(___(3)___) |> 
  head(5)

Blank (1)

  1. gender == "F"
  2. gender != "F"
  3. gender == "M"
  4. gender != "M"
  5. Both a and b
  6. Both a and c
  7. Both a and d
  8. Both b and c
  9. Both b and d
  10. Both c and d

g


Blank (2)

  1. name
  2. gender
  3. n
  4. name, n
  5. gender, n
  6. name, gender
  7. gender, name

a


Blank (3)

  1. n
  2. -n
  3. desc(n)
  4. Both b and c

d


Question 36

  • Likewise, using nyc_dogs_filtered from Question 34, we are creating the top5names_M data.frame:
name n
max 1341
charlie 1042
rocky 1020
buddy 840
teddy 745
  • The top5names_M data.frame provides the five most popular male dog names, displayed above.

Complete the code by filling in the blanks (1)-(2).

top5names_M <- nyc_dogs_filtered |> 
  filter(___(1)___) |> 
  count(___(2)___) |> 
  arrange(___(3)___) |> 
  head(5)

Blank (1)

  1. gender == "F"
  2. gender != "F"
  3. gender == "M"
  4. gender != "M"
  5. Both a and b
  6. Both a and c
  7. Both a and d
  8. Both b and c
  9. Both b and d
  10. Both c and d

h


Blank (2)

  1. name
  2. gender
  3. n
  4. name, n
  5. gender, n
  6. name, gender
  7. gender, name

a


Blank (3)

  1. n
  2. -n
  3. desc(n)
  4. Both b and c

d


Questions 37-40

The Nobel Prize in Economic Science in 2021 goes to David Card, Joshua Angrist and Guido Imbens, for their empirical contributions to labor economics, and for their methodological contributions to the analysis of causal relationships.

They have provided us with new insights about the labor market and shown what conclusions about cause and effect can be drawn from natural experiments. Their approach has spread to other fields and revolutionized empirical research.

For Questions 37-40, consider the following R packages and the data.frame, ak91_age, which comes from the 1980 US Census and covers men born 1930–1939, which is used by Joshua Angrist and Alan Krueger’s research article.

library(tidyverse)
library(skimr)
library(ggthemes)

ak91_age <- read_csv('https://bcdanl.github.io/data/ak91_ageW.csv')

The first 20 observations in the ak91_age data frame are displayed below:

QoB YoB YoBQ W Educ Q4
1 1930 1930.00 361.0922 12.28041 FALSE
1 1931 1931.00 365.8181 12.54043 FALSE
1 1932 1932.00 364.9678 12.53393 FALSE
1 1933 1933.00 362.1093 12.67319 FALSE
1 1934 1934.00 363.2739 12.64726 FALSE
1 1935 1935.00 357.7532 12.65091 FALSE
1 1936 1936.00 359.5803 12.74304 FALSE
1 1937 1937.00 362.5073 12.83230 FALSE
1 1938 1938.00 362.9918 12.93868 FALSE
1 1939 1939.00 360.0860 13.00299 FALSE
2 1930 1930.25 364.3105 12.42842 FALSE
2 1931 1931.25 365.2228 12.53105 FALSE
2 1932 1932.25 365.2356 12.60960 FALSE
2 1933 1933.25 365.2171 12.63471 FALSE
2 1934 1934.25 362.2778 12.72797 FALSE
2 1935 1935.25 360.1939 12.79693 FALSE
2 1936 1936.25 360.2046 12.81108 FALSE
2 1937 1937.25 360.7164 12.84405 FALSE
2 1938 1938.25 366.8558 13.00766 FALSE
2 1939 1939.25 365.9290 13.01340 FALSE
  • The ak91_age data frame is with 40 observations and 6 variables.

Description of Variables in ak91_age:

  • QoB: Quarter of birth
  • YoB: Year of birth (1930, 1931, …, 1939)
  • YoBQ: Year and quarter of birth (1930 Q1, 1930 Q2, …, 1939 Q4)
  • W: Wage per week
  • Educ: Years of education
  • Q4: TRUE if quarter of birth is 4; FALSE otherwise.


The followings are the summary of the ak91_age data.frame, including descriptive statistics for each variable.

Data summary
Name ak91_age
Number of rows 40
Number of columns 6
_______________________
Column type frequency:
logical 1
numeric 5
________________________
Group variables None

Variable type: logical

skim_variable n_missing mean count
Q4 0 0.25 FAL: 30, TRU: 10

Variable type: numeric

skim_variable n_missing mean sd p0 p25 p50 p75 p100
QoB 0 2.50 1.13 1.00 1.75 2.50 3.25 4.00
YoB 0 1934.50 2.91 1930.00 1932.00 1934.50 1937.00 1939.00
YoBQ 0 1934.88 2.92 1930.00 1932.44 1934.88 1937.31 1939.75
W 0 365.02 3.37 357.75 362.24 365.53 367.89 370.32
Educ 0 12.76 0.19 12.28 12.64 12.75 12.93 13.12

Question 37

Here we describe the quarterly trend of years of education. Complete the code by filling in the blanks (1)-(3).

ggplot(data = ak91_age, 
       mapping = aes(___(1)___,
                     ___(2)___ = Q4)) +
  ___(3)___ + 
  geom_point(size = 2.5) +
  scale_color_colorblind() +
  labs(x = "Year and quarter of birth",
       y = "Years of education")

Blank (1)

  1. x = YoBQ, y = Educ
  2. y = YoBQ, x = Educ
  3. x = YoB, y = Educ
  4. y = YoB, x = Educ
  5. Both a and b
  6. Both c and d

a


Blank (2)

  1. fill
  2. color
  3. Both a and b

b


Blank (3)

  1. geom_scatterplot
  2. geom_point
  3. geom_line
  4. geom_smooth
  5. geom_histogram
  6. geom_boxplot
  7. geom_bar
  8. geom_col

c


Question 38

Here we describe the quarterly trend of the base-10 log of wage per week. Complete the code by filling in the blanks (1)-(3).

ggplot(data = ak91_age, 
       mapping = aes(___(1)___,
                     ___(2)___ = Q4)) +
  ___(3)___ + 
  geom_point(size = 2.5) +
  scale_color_colorblind() +
  labs(x = "Year and quarter of birth",
       y = "Wage per week (in base-10 log)")

Blank (1)

  1. x = YoBQ, y = log(W)
  2. y = YoBQ, x = log(W)
  3. x = YoBQ, y = log10(W)
  4. y = YoBQ, x = log10(W)
  5. Both a and c
  6. Both b and d

c


Blank (2)

  1. fill
  2. color
  3. Both a and b

b


Blank (3)

  1. geom_scatterplot
  2. geom_point
  3. geom_line
  4. geom_smooth
  5. geom_histogram
  6. geom_boxplot
  7. geom_bar
  8. geom_col

c


Question 39

Here we describe how the distribution of the base-10 log of wage per week varies by Q4. Complete the code by filling in the blanks (1)-(3).

ggplot(data = ak91_age, 
       mapping = aes(___(1)___,
                     ___(2)___ = Q4)) +
  ___(3)___(show.legend = FALSE) +
  scale_fill_tableau() +
  labs(x = "Wage per week (in base-10 log)",
       y = "Born in the Fourth Quarter?")

Blank (1)

  1. x = YoBQ, y = log(W)
  2. y = YoBQ, x = log(W)
  3. x = YoBQ, y = log10(W)
  4. y = YoBQ, x = log10(W)
  5. x = Q4, y = log(W)
  6. y = Q4, x = log(W)
  7. x = Q4, y = log10(W)
  8. y = Q4, x = log10(W)
  9. Both a and c
  10. Both b and d
  11. Both e and g
  12. Both f and h

h


Blank (2)

  1. fill
  2. color
  3. Both a and b

a


Blank (3)

  1. geom_scatterplot
  2. geom_point
  3. geom_line
  4. geom_smooth
  5. geom_histogram
  6. geom_boxplot
  7. geom_bar
  8. geom_col

f


Question 40

Provide a data-driven narrative for the ak91_age data frame, incorporating insights from the visualizations created in Questions 37, 38, and 39.


Questions 41-44

For Questions 41-44, consider the following R packages and the data.frame, health_cust, which contains demographic information about individuals with or without health insurance.

library(tidyverse)
library(skimr)
library(ggthemes)

health_cust <- read_csv(
  'https://bcdanl.github.io/data/custdata_rev.csv'
)

The first 10 observations in the health_cust data frame are displayed below:

custid sex is_employed income marital_status housing_type
000006646_03 Male TRUE 22000 Never married Homeowner free and clear
000007827_01 Female NA 23200 Divorced/Separated Rented
000008359_04 Female TRUE 21000 Never married Homeowner with mortgage/loan
000008529_01 Female NA 37770 Widowed Homeowner free and clear
000008744_02 Male TRUE 39000 Divorced/Separated Rented
000011466_01 Male NA 11100 Married Homeowner free and clear
000015018_01 Female TRUE 25800 Married Rented
000017314_02 Female NA 34600 Married Homeowner free and clear
000017383_04 Female TRUE 25000 Never married Homeowner free and clear
000017554_02 Male TRUE 31200 Married Homeowner with mortgage/loan
custid recent_move num_vehicles age state_of_res gas_usage health_ins
000006646_03 FALSE 0 24 Alabama 210 FALSE
000007827_01 TRUE 0 82 Alabama 3 FALSE
000008359_04 FALSE 2 31 Alabama 40 FALSE
000008529_01 FALSE 1 93 Alabama 120 FALSE
000008744_02 FALSE 2 67 Alabama 3 FALSE
000011466_01 FALSE 2 76 Alabama 200 FALSE
000015018_01 FALSE 2 26 Alabama 3 TRUE
000017314_02 FALSE 2 73 Alabama 50 FALSE
000017383_04 FALSE 5 27 Alabama 3 FALSE
000017554_02 FALSE 3 54 Alabama 20 FALSE

Description of Variables in health_cust

  • custid: ID of customer
  • sex: Sex
  • is_employed: Employment status
    • NA: Unknown or not applicable
    • TRUE: Employed
    • FALSE: Unemployed
  • income: Income (in $)
  • marital_status: Marital status
  • housing_type: Housing type
  • recent_move:
    • TRUE: Recently moved
    • FALSE: Not recently moved
  • age: Age
  • state_of_res: State of residence (Alabama, Alaska, …, New York, …, Wyoming)
  • gas_usage: Gas usage
    • NA: Unknown or not applicable
    • 001: Included in rent or condo fee
    • 002: Included in electricity payment
    • 003: No charge or gas not used
    • 004-999: $4 to $999 (rounded and top-coded)
  • health_ins: Health insuarance status
    • TRUE: customer with health insuarance
    • FALSE: customer without health insuarance

The followings are the summary of the health_cust data.frame, including descriptive statistics for each variable.

Data summary
Name health_cust
Number of rows 73262
Number of columns 12
_______________________
Column type frequency:
character 5
logical 3
numeric 4
________________________
Group variables None

Variable type: character

skim_variable n_missing min max empty n_unique
custid 0 12 12 0 73262
sex 0 4 6 0 2
marital_status 0 7 18 0 4
housing_type 1720 6 28 0 4
state_of_res 0 4 20 0 51

Variable type: logical

skim_variable n_missing mean count
is_employed 25774 0.95 TRU: 45137, FAL: 2351
recent_move 1721 0.13 FAL: 62418, TRU: 9123
health_ins 0 0.10 FAL: 65955, TRU: 7307

Variable type: numeric

skim_variable n_missing mean sd p0 p25 p50 p75 p100
income 0 41764.15 58113.76 -6900 10700 26200 51700 1257000
num_vehicles 1720 2.07 1.17 0 1 2 3 6
age 0 49.16 18.08 0 34 48 62 120
gas_usage 1720 41.17 63.05 1 3 10 60 570

Question 41

Here we describe how the distribution of health_ins varies by state of residence and employment status using the health_cust data.frame. Complete the code by filling in the blanks (1)-(4).

ggplot(data = health_cust |> filter(!is.na(is_employed)),
       mapping = aes(___(1)___, 
                     fill = ___(2)___)) +
  ___(3)___ +
  ___(4)___(~is_employed) +
  labs(y = "", x= "Proportion") +
  scale_fill_tableau()

Blank (1)

  1. x = health_ins
  2. x = state_of_res
  3. x = Proportion
  4. y = health_ins
  5. y = state_of_res
  6. y = Proportion

e


Blank (2)

  1. health_ins
  2. state_of_res
  3. Proportion

a


Blank (3)

  1. geom_bar(position = "stack")
  2. geom_col(position = "stack")
  3. geom_bar(position = "fill")
  4. geom_col(position = "fill")
  5. geom_bar(position = "dodge")
  6. geom_col(position = "dodge")

c


Blank (4)

Answer: ________________________________________

facet_wrap


Question 42

Here we describe how the distribution of marital_status varies by housing_type using the health_cust data.frame. Complete the code by filling in the blanks (1)-(4).

ggplot(data = health_cust |> filter(!is.na(housing_type)),
       mapping = aes(___(1)___, 
                     ___(2)___)) +
  ___(3)___(show.legend = FALSE) +
  ___(4)___(~housing_type) +
  labs(y = "")

Blank (1)

  1. x = marital_status
  2. y = marital_status

b


Blank (2)

  1. x = prop
  2. x = prop, group = 1
  3. Both a and b
  4. y = prop
  5. y = prop, group = 1
  6. Both d and e
  7. x = after_stat(prop)
  8. x = after_stat(prop), group = 1
  9. Both g and h
  10. y = after_stat(prop)
  11. y = after_stat(prop), group = 1
  12. Both j and k

h


Blank (3)

  1. geom_bar()
  2. geom_col()
  3. Both a and b

a


Blank (4)

Answer: ________________________________________

facet_wrap


Question 43

Here we describe how the relationship between age and income varies by health_ins using the health_cust data.frame. Note that the new geometric object geom_hex() divides the plane into regular hexagons, counts the number of observations in each hexagon, and then maps the number of observations to the hexagon fill.

Complete the code by filling in the blanks (1)-(4).

# Considering 
  # income level between $0 and $250,000
  # age between 20 and 70
ggplot(data = health_cust |> filter(income >= 0 & income <= 2.5*10^5,
                                    age >= 20 & age <= 70),
       mapping = aes(___(1)___)) +
  geom_hex() + # hexbin plot: dividing the plot area into hexagonal bins
  ___(2)___ +
  ___(3)___(~health_ins) +
  scale_fill_viridis_c() # for hexbin color

Blank (1)

  1. x = income, y = age
  2. x = age, y = income

b


Blank (2)

  1. geom_smooth()
  2. geom_smooth(method = "lm")
  3. Both a and b

a


Blank (3)

Answer: ________________________________________

facet_wrap


Question 44

Describe how the overall relationship between age and income varies by health_ins.


Section 4. Short Answer

Question 45

For each question in Homework 5, briefly describe the task you are required to complete.


Question 46

What is clutter in data visualization, and why is it important to reduce it? Provide at least two practical tips for minimizing clutter in visualizations.

  • Clutter: Visual elements that occupy space but do not improve understanding

  • Clutter makes information harder to process and can confuse the viewer

    • Less clutter = clearer message, more focused audience

  • Tips

    • Avoid having the data all skewed to one side or the other of your graph.
    • Avoid too many superimposed elements, such as too many curves (>4) in the same graphing space.


Question 47

Describe the two phases of training a large language model (LLM): Pre-training and Fine-tuning. What is the primary objective of each phase?


Question 48

Compare supervised learning and unsupervised learning. Give one example of a business application for each and explain why labeled data is central to one but not the other.


Question 49

When is it appropriate to treat integer-valued data as if it were continuous? Give one example of an integer variable for which this is reasonable.


Question 50

Identify two situations where pie charts are not a suitable alternative to bar charts.

  1. Pie charts work well only if you only have a few categories—four max.

  2. Pie charts work well if the goal is to emphasize simple fractions (e.g., 25%, 50%, or 75%).

  3. Pie charts are not the best choice if you want audiences to compare the size of shares.

  4. Pie charts are not the best choice if you want audiences to compare the distribution across categories.


Back to top