Lecture 1
Dr. Robin Donatello
Office: Holt 202 Office hours:
Please share with at least two classmates…
Data science is an exciting discipline that allows you to turn raw data into understanding, insight, and knowledge.
We’re going to learn to do this in a modern and tidy way – more on that later!
This is a course on introduction to data science, with an emphasis on statistical thinking.
You took a survey, built in Google Forms:
We want to explore that data and get to know you!
More on what packages are on Thursday, but in a nutshell “get your tools out of the toolbox”:
library(tidyverse) # for data wrangling and visualization
library(scales) # for better axis labels
library(tidytext) # for handling text data
library(ggwordcloud) # for the learn-best word cloud
library(zipcodeR) # for the hometown map
library(maps) # also for mapping
library(ggpubr) # for better density plot
We could download the data and save it as a csv file on our computer, edit out the student names, name it survey-anonymized.csv and save it in a folder called data. Then we would import it into R using the read_csv() function:
Rows: 20 Columns: 7
── Column specification ──────────────────────────────────────────────
Delimiter: ","
chr (6): zipcode, stats_experience, programming_experience, progra...
dbl (1): commute_minutes
ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.Alternatively, we can authorize R to directly read the Google Sheet that holds the responses from the Google Form you filled out.
survey <- read_sheet("https://docs.google.com/spreadsheets/d/1R4_afqSB6KXD9HwfYP-wjMqvwxiDTGEMATmWmkqqXig")survey# A tibble: 20 × 7
zipcode stats_experience programming_experience
<chr> <chr> <chr>
1 95928 Yes, I have taken a high school sta… A little — I've writt…
2 95818 No, I have not taken any statistics… None
3 95926 Yes, I have taken another college c… Some — I've worked on…
4 90210 No, I have not taken any statistics… A lot — I use program…
5 95965 Yes, I have taken a high school sta… None
6 96001 No, I have not taken any statistics… A little — I've writt…
7 97201 Yes, I have taken another college c… Some — I've worked on…
8 95973 No, I have not taken any statistics… A lot — I use program…
9 93701 Yes, I have taken a high school sta… A little — I've writt…
10 95991 No, I have not taken any statistics… None
11 95926 Yes, I have taken another college c… Some — I've worked on…
12 00000 No, I have not taken any statistics… A little — I've writt…
13 95928 Yes, I have taken a high school sta… None
14 90210 No, I have not taken any statistics… A lot — I use program…
15 95814 Yes, I have taken another college c… Some — I've worked on…
16 95965 No, I have not taken any statistics… A little — I've writt…
17 85001 Yes, I have taken a high school sta… None
18 92101 No, I have not taken any statistics… Some — I've worked on…
19 95973 Yes, I have taken another college c… A lot — I use program…
20 96001 No, I have not taken any statistics… A little — I've writt…
# ℹ 4 more variables: programming_languages <chr>,
# data_interests <chr>, learn_best <chr>, commute_minutes <dbl>We asked you the following multiple-choice question where you could only pick one option:
Have you taken any statistics courses before?
One way to make sense of data collected via a question like this is to visualize it.
survey |>
count(stats_experience) |>
mutate(prop = n / sum(n)) |>
ggplot(aes(y = stats_experience, x = prop))+
geom_col(show.legend = FALSE) +
scale_y_discrete(labels = label_wrap(20)) +
scale_x_continuous(labels = percent_format(accuracy = 1), breaks = c(0, 0.25, 0.5)) +
labs(
title = "Prior statistics experience among DATA 385 students",
y = NULL,
x = "Count"
) +
labs(
caption = "Data are self-reported, collected from DATA 385 students on August 24, 2026."
)
We also asked you the following multiple-choice question where you could only pick one option:
How much experience do you have with programming?
survey |>
count(programming_experience) |>
mutate(prop = n / sum(n)) |>
ggplot(aes(y = fct_reorder(programming_experience, prop), x = prop, fill = prop))+
geom_col(show.legend = FALSE) +
scale_y_discrete(labels = label_wrap(25)) +
scale_x_continuous(labels = percent_format(accuracy = 1), breaks = c(0, 0.1, 0.2, 0.3, 0.4)) +
scale_fill_viridis_c(option = "E") +
labs(
title = "Prior programming experience among DATA 385 students",
y = NULL,
x = "Count"
) +
labs(
caption = "Data are self-reported, collected from DATA 385 students on August 24, 2026."
) +
theme_minimal(base_size = 16)
We also asked you the following multiple-choice question where you could as many options as you liked:
What types of data interest you?
survey |>
select(data_interests)# A tibble: 20 × 1
data_interests
<chr>
1 Sports, Entertainment (e.g., books, movies, music)
2 Crime, Politics
3 Health (e.g., social determinants of health, medical), Environment…
4 Economics, Sports, Politics
5 Entertainment (e.g., books, movies, music), Education
6 No preference
7 Environment/Climate
8 Sports, Entertainment (e.g., books, movies, music), Other
9 Health (e.g., social determinants of health, medical), Education, …
10 Politics, Economics
11 No preference
12 Sports, Environment/Climate
13 Entertainment (e.g., books, movies, music), Other
14 Economics, Crime, Politics
15 Health (e.g., social determinants of health, medical), Education
16 Sports, Politics, Other
17 Entertainment (e.g., books, movies, music), Environment/Climate, E…
18 No preference
19 Crime, Health (e.g., social determinants of health, medical), Spor…
20 Entertainment (e.g., books, movies, music), Other Before we can visualize this variable, we need to tidy and transform it.
survey |>
# remove text in parentheses
mutate(data_interests = str_remove_all(data_interests, "\\s*\\(.*?\\)")) |>
# separate into multiple rows for each interest, using comma as delimiter
separate_longer_delim(data_interests, delim = ",") |>
count(data_interests, sort = TRUE) |>
mutate(prop = n / sum(n))# A tibble: 17 × 3
data_interests n prop
<chr> <int> <dbl>
1 " Education" 4 0.0930
2 " Other" 4 0.0930
3 " Politics" 4 0.0930
4 "Entertainment" 4 0.0930
5 "Sports" 4 0.0930
6 " Environment/Climate" 3 0.0698
7 "Health" 3 0.0698
8 "No preference" 3 0.0698
9 " Crime" 2 0.0465
10 " Entertainment" 2 0.0465
11 " Sports" 2 0.0465
12 "Crime" 2 0.0465
13 "Economics" 2 0.0465
14 " Economics" 1 0.0233
15 " Health" 1 0.0233
16 "Environment/Climate" 1 0.0233
17 "Politics" 1 0.0233survey |>
mutate(data_interests = str_remove_all(data_interests, "\\s*\\(.*?\\)")) |>
separate_longer_delim(data_interests, delim = ",") |>
count(data_interests, sort = TRUE) |>
mutate(prop = n / sum(n)) |>
filter(!is.na(data_interests)) |>
ggplot(aes(y = fct_reorder(data_interests, prop), x = prop, fill = prop)) +
geom_col(show.legend = FALSE) +
scale_y_discrete(labels = label_wrap(20)) +
scale_x_continuous(labels = percent_format(accuracy = 1)) +
scale_fill_distiller(palette = "YlOrBr") +
labs(
title = "Data interests among DATA 385 students",
y = NULL,
x = "Count",
caption = "Data are self-reported, collected from DATA 385 students on August 24, 2026."
) +
theme_minimal(base_size = 16)
We also asked you the following open-ended, numeric question:
How long does it take you to commute from home to this classroom in minutes?
survey |>
ggdensity(
x = "commute_minutes",
fill = "lightblue",
color = "darkblue",
add_density = TRUE,
add = "mean",
rug = TRUE,
title = "Commute time to class among DATA 385 students",
xlab = "Commute time (minutes)",
ylab = "Density"
) +
labs(caption = "Data are self-reported, collected from DATA 385 students on August 24, 2026.")
We also asked you the following open-ended question:
How do you learn best?
survey |>
select(learn_best)# A tibble: 20 × 1
learn_best
<chr>
1 I learn best by doing hands-on practice rather than just reading o…
2 Visual explanations with diagrams help me a lot.
3 I retain things best when I can teach them back to someone else.
4 I like working through lots of small examples until it clicks.
5 Repetition and flashcards work well for me.
6 I need to see a real example before the concept makes sense.
7 I learn best in study groups where we talk through problems out lo…
8 Building small projects on my own is how it really sinks in for me.
9 I do best with step-by-step written instructions I can follow alon…
10 Lecture plus notes works fine for me but I need to rewrite my not…
11 I'm a visual learner - charts and diagrams over text every time.
12 I learn best by breaking things and figuring out how to fix them.
13 Podcasts and videos stick with me better than reading textbooks.
14 I like diving straight into the code and learning by debugging err…
15 Office hours and asking questions one-on-one really help me.
16 I learn best from worked examples I can compare my own work agains…
17 I need to write things out by hand to actually learn them.
18 Group projects where I can bounce ideas off others work best for m…
19 I'm most productive with a clear checklist and deadlines to work t…
20 Trial and error - I just start messing with things until they work.We can use text mining techniques, like tokenizing to words to explore this open-ended question:
We can also tokenize to bigrams (pairs of words):
survey |>
select(learn_best) |>
tidytext::unnest_tokens(bigrams, learn_best, token = "ngrams", n = 2) |>
count(bigrams, sort = TRUE) |>
filter(n > 1) |>
print(n = Inf)# A tibble: 10 × 2
bigrams n
<chr> <int>
1 for me 4
2 i can 4
3 i learn 4
4 learn best 4
5 i need 3
6 need to 3
7 best by 2
8 help me 2
9 i like 2
10 my own 2survey |>
select(learn_best) |>
unnest_tokens(word, learn_best) |>
anti_join(stop_words, by = "word") |>
count(word, sort = TRUE) |>
ggplot(aes(label = word, size = n)) +
ggwordcloud::geom_text_wordcloud(area_corr = TRUE) +
scale_size_area(max_size = 24) +
labs(title = "How DATA 385 students learn best") +
theme_minimal(base_size = 16)
We also asked you the following question:
What is the zipcode of your hometown?
If you are from out of country, please enter 00000.
survey_locations <- survey |>
mutate(zipcode = str_pad(zipcode, width = 5, side = "left", pad = "0")) |>
left_join(zip_code_db, by = "zipcode")
survey_locations |>
filter(!is.na(lat), zipcode != "00000") |>
ggplot(aes(x = lng, y = lat)) +
borders("state", colour = "gray85", fill = "gray95") +
geom_point(color = "#2C4A5E", size = 3, alpha = 0.7) +
coord_quickmap() +
theme_void() +
labs(
title = "Where DATA 385 students call home",
caption = "Data are self-reported, collected from DATA 385 students on August 24, 2026."
)Warning: `borders()` was deprecated in ggplot2 4.0.0.
ℹ Please use `annotation_borders()` instead.
Note: rows with zipcode 00000 (international students) are excluded from the map — worth calling out as a group separately rather than silently dropping them.
We also asked you the following open-ended question:
If you’ve programmed before, which languages have you used, and how comfortable do you feel with each? If you haven’t programmed before, please leave this question blank.
And the answers are non-trivial to tidy up, e.g.,
survey |>
select(programming_languages)# A tibble: 20 × 1
programming_languages
<chr>
1 Python - just a intro course
2 <NA>
3 Python - pretty comfortable, SQL - a little
4 Java - very comfortable, Python - very comfortable, C++ - comforta…
5 <NA>
6 HTML/CSS - beginner
7 Python - comfortable, R - a little
8 Python - very comfortable, JavaScript - comfortable, Java - a litt…
9 Python - a little from a summer camp
10 <NA>
11 Python - comfortable, MATLAB - a little
12 C - a little in high school
13 <NA>
14 Python - very comfortable, C++ - very comfortable, Rust - a little
15 R - comfortable, Python - a little
16 Python - one intro class
17 <NA>
18 JavaScript - comfortable, Python - a little
19 Python - very comfortable, SQL - comfortable, Java - comfortable
20 Scratch - from a middle school class Prompt:
Summarize the following responses to the question “If you’ve programmed before, which languages have you used, and how comfortable do you feel with each? If you haven’t programmed before, please leave this question blank.”. Write your response in a short paragraph.
Response:
Python is by far the most common language, ranging from a little exposure (summer camp, one intro course) to very comfortable. Several students also have experience with a second language — Java, C++, and JavaScript each show up a few times, generally at moderate-to-strong comfort levels — while others mention lighter exposure to SQL, R, C, MATLAB, Rust, or HTML/CSS. A couple of students noted only minimal or early exposure (e.g., Scratch from middle school, C in high school). Overall, the class has a wide range of backgrounds, from complete beginners with just one intro course to students who are very comfortable across multiple languages.
No. They can be misleading, inaccurate, hallucinate or in general sound more confident than they are. Humans should always stay “in the loop”. Hence – the interdisciplinary approach!
Most videos linked throughout this course come from Duke’s DS in a Box / STA 199 courses, recorded around 2020. A few things to watch for:
This is a flipped classroom: content prep happens before class, and class time is for code-alongs and collaborative work.
| Category | Percentage |
|---|---|
| Labs | 25% |
| Homework/Reading Responses | 20% |
| Exams | 30% |
| Quizzes | 10% |
| Project | 10% |
| Participation | 5% |
Standard +/– letter scale: A (90–100), B (80–89), C (70–79), D (60–69), F (below 60)
See course syllabus for full grading policy details.