In this project, we will explore 100+ years of US Baby Names. Although this dataset only has 4 columns, there are so many interesting questions one could explore. While the possibilities are endless, here are five interesting questions for you to explore.
You need to complete at least the first three questions. The last two questions are optional.
Let us start by importing the data from
data/names.csv.gz. You can use the read_csv
function from the readr package to do this.
# Load the tidyverse package
library(tidyverse)
# Read the csv file `file_name_names` as data frame `tbl_names`
file_name_names <- here::here("data/names.csv.gz")
tbl_names <- readr::read_csv(
file = file_name_names,
show_col_types = FALSE
)
# Print `tbl_names`
tbl_names#> # A tibble: 2,052,781 × 4
#> year name sex nb_births
#> <dbl> <chr> <chr> <dbl>
#> 1 1880 Mary F 7065
#> 2 1880 Anna F 2604
#> 3 1880 Emma F 2003
#> 4 1880 Elizabeth F 1939
#> 5 1880 Minnie F 1746
#> 6 1880 Margaret F 1578
#> 7 1880 Ida F 1472
#> 8 1880 Alice F 1414
#> 9 1880 Bertha F 1320
#> 10 1880 Sarah F 1288
#> # ℹ 2,052,771 more rowsOne of the first things we want to do is to understand naming trends. Let us start by figuring out the top five most popular male and female names for this decade (born 2011 and after, but before 2021). Do you want to make any guesses?
The first step is always to transform the data into a form that is
easy to visualize. If we create a table with sex,
name and nb_births for the top 5 names for
each sex, then we can turn it into a column plot using
ggplot2. Let us get the five most popular male and female
names of the decade starting in 2011.
tbl_names_popular = tbl_names |>
# Keep ROWS for year > 2010 and <= 2020
filter(year > 2010, year <= 2020) |>
# Group by sex and name
group_by(sex, name) |>
# Summarize the number of births
summarize(
nb_births = sum(nb_births),
.groups = "drop"
) |>
# Group by sex
group_by(sex) |>
# For each sex, keep the top 5 rows by number of births
slice_max(nb_births, n = 5)
tbl_names_popular#> # A tibble: 10 × 3
#> # Groups: sex [2]
#> sex name nb_births
#> <chr> <chr> <dbl>
#> 1 F Emma 193138
#> 2 F Olivia 184966
#> 3 F Sophia 173341
#> 4 F Isabella 159570
#> 5 F Ava 153414
#> 6 M Noah 184977
#> 7 M Liam 182646
#> 8 M William 155326
#> 9 M Mason 152944
#> 10 M Jacob 150145Let us now visualize this data as a horizontal bar plot of number of births by name faceted by sex.
Here, when thinking about how we might label this table, look back to the statement of the question. Make sure to use this information in a way so that the title, subtitle, and caption provide your reader with concise, descriptive language so that viewers can quickly grasp the plot’s purpose and findings.
tbl_names_popular |>
# Reorder the names by number of births
mutate(name = fct_reorder(name, nb_births)) |>
# Initialize a ggplot for name vs. nb_births
ggplot(aes(x = nb_births, y = name)) +
# Add a column plot layer
geom_col() +
# Facet the plots by sex
facet_wrap(~ sex, scales = "free_y") +
# Add labels (title, subtitle, caption, x, y)
labs(
title = 'The Most Popular US Baby Names',
subtitle = 'There are two most popular names',
caption = 'Source: SSA',
x = 'Births ',
y = 'Name'
) +
# Fix the x-axis scale
scale_x_continuous(
labels = scales::unit_format(scale = 1e-3, unit = "K"),
expand = c(0, 0),
) +
# Move the plot title to top left
theme(
plot.title.position = 'plot'
)
tbl_names_popular#> # A tibble: 10 × 3
#> # Groups: sex [2]
#> sex name nb_births
#> <chr> <chr> <dbl>
#> 1 F Emma 193138
#> 2 F Olivia 184966
#> 3 F Sophia 173341
#> 4 F Isabella 159570
#> 5 F Ava 153414
#> 6 M Noah 184977
#> 7 M Liam 182646
#> 8 M William 155326
#> 9 M Mason 152944
#> 10 M Jacob 150145Consider the following two names Elizabeth and
Deneen. Elizabeth is a fairly stable name,
while Deneen is not, when one looks at trends. An
interesting question to ask would be what are the top 5 stable and top 5
trendiest names. A stable name is one whose numbers across
years don’t vary drastically, while a trendy name is one whose
popularity peaks for a short period and then dies down.
There are many ways to capture trendiness. A simple measure would be to look at the maximum number of births for a name, which is then adjusted to account for differences in the total number of births across years. This process of adjustment is called normalizing, and it allows for a fair comparison of the trendiness of names across different time periods. A trendy name would have a high value after being normalized in this way.
Let us use this idea to figure out the top 10 trendy names in this data set. Let us use a cutoff of at least a 1000 births across all the years, so we don’t pick up low-volume names. Feel free to experiment with this threshold as you see fit.
We need to transform the data into a table that consists of the
columns sex, name,
nb_births_total, nb_births_max and
trendiness. We compute nb_births_total as the
total number of births across all years, and nb_births_max
as the maximum number of births for a given name across all years.
Finally, we compute trendiness as a ratio of these two numbers. Follow
the recipe outlined below to carry out the transformation.
tbl_names_popular_trendy = tbl_names |>
# Group by sex and name
group_by(sex, name) |>
# Summarize total number of births and max births in a year
summarize(
nb_births_total = sum(nb_births),
nb_births_max = max(nb_births),
.groups = "drop"
) |>
# Filter for names with at least 10000 births
filter(nb_births_total > 10000 ) |>
# Add a column for trendiness computed as ratio of max to total
mutate(trendiness = nb_births_max / nb_births_total) |>
# Group by sex
group_by(sex) |>
# Slice top 5 rows by trendiness for each group
slice_max(trendiness, n = 5)
tbl_names_popular_trendy#> # A tibble: 10 × 5
#> # Groups: sex [2]
#> sex name nb_births_total nb_births_max trendiness
#> <chr> <chr> <dbl> <dbl> <dbl>
#> 1 F Katina 11284 2745 0.243
#> 2 F Ashanti 12141 2945 0.243
#> 3 F Marquita 11010 2543 0.231
#> 4 F Everleigh 12738 2814 0.221
#> 5 F Miley 12865 2649 0.206
#> 6 M Luka 14393 3197 0.222
#> 7 M Atlas 11869 2523 0.213
#> 8 M Jase 22062 4552 0.206
#> 9 M Legend 16153 3152 0.195
#> 10 M Jayceon 11606 2013 0.173| Note |
| There are always several different approaches to the same transformation. When it comes to learning, there’s no one-size-fits-all solution. One way to make your learning stick is by trying to solve a problem in different ways. Even if you don’t get to the final answer, you will learn a lot by exploring different approaches. |
Let us write a function that will accept a name as an argument and return a plot with trends in number of births. We can then use this function to plot the trends for different names.
plot_trends_in_name <- function(my_name) {
tbl_names |>
# Filter for name = my_name
filter(name == my_name) |>
# Initialize a ggplot of `nb_births` vs. `year` colored by `sex`
ggplot(aes(x = year, y = nb_births, color = sex)) +
# Add a line layer
geom_line() +
# Add labels (title, x, y)
labs(
title = glue::glue("Babies named {my_name} across the years!"),
x = 'Year',
y = 'Number of births'
) +
# Update plot theme
theme(plot.title.position = "plot")
}
plot_trends_in_name("Steve")
plot_trends_in_name("Barbara")
Have you ever wondered why some letters seem to be more prevalent in names than others? In this question, you will embark on a journey to uncover the reasons behind the popularity of specific letters in names. This investigation will lead you to interesting insights about how the popularity of letters in names has changed over time and the potential factors that have influenced these trends.
Let us start by transforming the data and adding two columns, one for
first_letter and one for last_letter. You can
use the str_sub() function from the stringr
package to do this.
tbl_names = tbl_names |>
# Add NEW column first_letter by extracting `first_letter` from name using `str_sub`
mutate(first_letter = str_sub(name, 1, 1)) |>
# Add NEW column last_letter by extracting `last_letter` from name using `str_sub`
mutate(last_letter = str_sub(name, -1, -1)) |>
# UPDATE column `last_letter` to upper case using `str_to_upper`
mutate(last_letter = str_to_upper(last_letter))
tbl_names#> # A tibble: 2,052,781 × 6
#> year name sex nb_births first_letter last_letter
#> <dbl> <chr> <chr> <dbl> <chr> <chr>
#> 1 1880 Mary F 7065 M Y
#> 2 1880 Anna F 2604 A A
#> 3 1880 Emma F 2003 E A
#> 4 1880 Elizabeth F 1939 E H
#> 5 1880 Minnie F 1746 M E
#> 6 1880 Margaret F 1578 M T
#> 7 1880 Ida F 1472 I A
#> 8 1880 Alice F 1414 A E
#> 9 1880 Bertha F 1320 B A
#> 10 1880 Sarah F 1288 S H
#> # ℹ 2,052,771 more rowsBegin by computing the distribution of births across year and sex by first letter of a name.
tbl_names_by_letter = tbl_names |>
# Group by year, sex and first_letter
group_by(sex, year, first_letter) |>
# Summarize total number of births, drop the grouping
summarise(nb_births = sum(nb_births), .groups = "drop") |>
# Group by year and sex
group_by(year, sex) |>
# Add NEW column pct_births by dividing nb_births by sum(nb_births)
mutate(pct_births = nb_births / sum(nb_births))
tbl_names_by_letter#> # A tibble: 7,313 × 5
#> # Groups: year, sex [284]
#> sex year first_letter nb_births pct_births
#> <chr> <dbl> <chr> <dbl> <dbl>
#> 1 F 1880 A 9334 0.103
#> 2 F 1880 B 3876 0.0426
#> 3 F 1880 C 5868 0.0645
#> 4 F 1880 D 2218 0.0244
#> 5 F 1880 E 11444 0.126
#> 6 F 1880 F 2957 0.0325
#> 7 F 1880 G 2463 0.0271
#> 8 F 1880 H 2743 0.0301
#> 9 F 1880 I 2480 0.0273
#> 10 F 1880 J 3800 0.0418
#> # ℹ 7,303 more rowsVisualize the distribution of births by first letter for the year 2020, faceted by sex.
tbl_names_by_letter |>
# Filter for the year 2020
filter(year == 2020) |>
# Initialize a ggplot of pct_births vs. first_letter
ggplot(aes(x = first_letter, y = pct_births)) +
# Add a column layer using `geom_col()`
geom_col() +
# Facet wrap plot by sex
facet_wrap(~ sex) +
# Add labels (title, subtitle, x, y)
labs(
title = 'Most Female Baby name starts with A',
x = 'Baby names first letter',
y = 'Birth %'
) +
# Fix scales of y axis
scale_y_continuous(
expand = c(0, 0),
labels = scales::percent_format(accuracy = 1L)
) +
# Update plotting theme
theme(
plot.title.position = "plot",
axis.ticks.x = element_blank(),
panel.grid.major.x = element_blank()
)
Write a function that plot trends in the percentage of births for all names starting with a specific first letter.
plot_trends_in_letter <- function(my_letter) {
tbl_names_by_letter |>
# Filter for first_letter = my_letter
filter(first_letter == my_letter) |>
# Initialize a ggplot of pct_births vs. year colored by sex
ggplot(aes(x = year, y = pct_births, color = sex)) +
# Add a line layer
geom_line() +
# Add labels (title, subtitle, caption, x, y)
labs(
title = glue::glue("Trends in Names beginning with {my_letter}"),
subtitle = "Trends in Names",
caption = "Source: SSA",
x = "Year",
y = 'Birth %'
) +
# Update y-axis scales to display percentages
scale_y_continuous(labels = scales::percent_format()) +
# Update theme
theme(plot.title.position = "plot")
}
plot_trends_in_letter("S")
| [BONUS] |
How do these plots change when you focus on the
last_letter instead of the first_letter. Feel
free to go back and change first_letter to
last_letter and see what you find. Try to make an educated
guess before you run the code, so you can see if your intuition matches
up! |
Are you ready to explore the fascinating realm of letter combinations in names? This question will guide you through the process of analyzing the joint distribution of births by first and last letters. By examining these intriguing patterns, you’ll be able to unveil the most popular letter combinations and how they have evolved over the years.
tbl_names_by_first_and_last_letter = tbl_names |>
# Filter for sex = "F"
filter(sex == "F") |>
# Group by `first_letter`, `last_letter`, and `year`
group_by(first_letter,last_letter,year) |>
# Summarize total number of births
summarise(nb_births = sum(nb_births)) |>
# Group by `year`
group_by(year) |>
# Add NEW column pct_births by dividing nb_births by sum(nb_births)
mutate(pct_births = nb_births / sum(nb_births)) |>
# Ungroup data
ungroup()#> `summarise()` has grouped output by 'first_letter', 'last_letter'. You can
#> override using the `.groups` argument.tbl_names_by_first_and_last_letter#> # A tibble: 43,579 × 5
#> first_letter last_letter year nb_births pct_births
#> <chr> <chr> <dbl> <dbl> <dbl>
#> 1 A A 1880 4784 0.0526
#> 2 A A 1881 4966 0.0540
#> 3 A A 1882 5728 0.0531
#> 4 A A 1883 6051 0.0539
#> 5 A A 1884 7038 0.0546
#> 6 A A 1885 7203 0.0541
#> 7 A A 1886 7785 0.0539
#> 8 A A 1887 7814 0.0535
#> 9 A A 1888 9444 0.0529
#> 10 A A 1889 9365 0.0525
#> # ℹ 43,569 more rowsNow, you will visualize the distribution of pct_births
by last_letter and first_letter by plotting a
heatmap of the percentage of births by first letter and last letter for
the year 2021.
tbl_names_by_first_and_last_letter |>
# Filter for the year 2021
filter (year == 2021) |>
# Initialize a ggplot of last_letter vs. first_letter
ggplot(aes(x = last_letter, y = first_letter)) +
# Add a `geom_tile` layer with fill mapped to pct_births
geom_tile(aes(fill = pct_births)) +
# Add labels (title, subtitle, x, y, fill)
labs(
title = 'Distribution of First Name and Last Time in 2021',
subtile = 'AA is the most popular combination',
x = 'last letter',
y = 'first letter ',
fill = 'Birth %'
) +
# Update fill scale to use Viridis colors
scale_fill_viridis_b(direction = -1) +
# Update plotting theme
theme(
plot.title.position = "plot",
panel.grid = element_blank(),
axis.ticks = element_blank()
)
Do certain combinations of vowels and consonants tend to appear more often in names? Are there any notable changes in these patterns over the years? In this question, we’ll explore the fascinating world of vowel and consonant usage in names across time. This can help us understand how the structure of names has evolved and what factors may have influenced these changes. By diving into these linguistic aspects, you’ll gain a greater appreciation for the intricacies and diversity of names in our dataset. Let’s dive in and uncover the trends in the usage of vowels and consonants!
In this step, you will create a function to identify whether a letter is a vowel or a consonant. Then, you will use this function to categorize the first and last letters in names as either vowels or consonants. After that, you will group the data by sex, year, and letter type (vowel or consonant) to calculate the percentage of births for each combination of first and last letter types.
get_letter_type <- function(letter) {
VOWELS <- c("A", "E", "I", "O", "U")
ifelse(letter %in% VOWELS, 'vowel', 'consonant')
}
tbl_names_vowel_consonant <- tbl_names |>
# Add NEW column named `first_letter_type`
mutate(first_letter_type = get_letter_type(first_letter)) |>
# Add NEW column named `last_letter_type`
mutate(last_letter_type = get_letter_type(last_letter)) |>
# Group by `sex`, `year`, `first_letter_type` and `last_letter_type`
group_by(sex, year, first_letter_type, last_letter_type) |>
# Summarize the total number of births
summarize(nb_births = sum(nb_births)) |>
# Group by `sex` and` `year`
group_by(sex, year) |>
# Add NEW column with `pct_births` calculated as `nb_births / sum(nb_births)`
mutate(pct_births = nb_births/sum(nb_births)) |>
# Ungroup the data
ungroup() |>
# Unite `first_letter_type` and `last_letter_type` into a NEW column named `first_last`
unite("first_last", first_letter_type,last_letter_type)#> `summarise()` has grouped output by 'sex', 'year', 'first_letter_type'. You can
#> override using the `.groups` argument.tbl_names_vowel_consonant#> # A tibble: 1,136 × 5
#> sex year first_last nb_births pct_births
#> <chr> <dbl> <chr> <dbl> <dbl>
#> 1 F 1880 consonant_consonant 19988 0.220
#> 2 F 1880 consonant_vowel 46765 0.514
#> 3 F 1880 vowel_consonant 5708 0.0627
#> 4 F 1880 vowel_vowel 18533 0.204
#> 5 F 1881 consonant_consonant 20069 0.218
#> 6 F 1881 consonant_vowel 47287 0.514
#> 7 F 1881 vowel_consonant 5669 0.0617
#> 8 F 1881 vowel_vowel 18928 0.206
#> 9 F 1882 consonant_consonant 23561 0.218
#> 10 F 1882 consonant_vowel 55449 0.514
#> # ℹ 1,126 more rowsNow, you will create a visualization to display the trends in the usage of vowels and consonants in names over time. The visualization will show the percentage of births by the combination of first and last letter types, separately for each sex.
tbl_names_vowel_consonant |>
# Reorder `first_last` by the median `pct_births`
mutate(first_last = fct_reorder(first_last, pct_births, median)) |>
# Initialize a ggplot of `pct_births` vs. `year`
ggplot(aes(x = year, y = pct_births)) +
# Add an area layer with fill = first_last
geom_area(aes(fill = first_last)) +
# Facet wrap plot by `sex`
facet_wrap(~ sex) +
# Add labels (title, subtitle, caption, x, y)
labs(
title = " Consonants and vowels usage in baby names",
subtitle = "Male babies has most consonants in their name",
caption = "Source: SSA",
x = "year",
y = "Births % "
) +
# Clean up x and y axis scales
scale_x_continuous(
expand = c(0, 0)
) +
scale_y_continuous(
expand = c(0, 0),
labels = scales::percent_format()
) +
# Use Viridis colors for fill
scale_fill_viridis_d() +
# Update plotting theme
theme(
plot.title.position = 'plot',
legend.position = 'bottom'
)
The fct_reorder() function is a useful tool for
reordering the levels of a factor based on the values of another
variable. This function can be particularly helpful when creating
visualizations, as it allows you to display data in a more organized and
informative manner. In this exercise, fct_reorder() is used
to reorder the levels of the first_last factor according to the median
of the pct_births variable. By doing so, the resulting plot
will show the trends in vowels and consonants usage in a more structured
way, making it easier to identify and interpret the patterns.
Congratulations 🥳🎉 on completing your first project. Follow the instructions below to submit it for review.
Click on the Knit button in the menu on top
This will render the document to a markdown file with the same name.
Commit both Rmd and md documents to github
Navigate to the git tab, check both the
Rmd and md files for the project, and click on
the Commit button.
Add a meaningful commit message and click on the Commit button to push your changes to your github repository.
You will need to make sure that you give gitpod permissions to read/write to your github repository. Instructions on how to this can be found on the Corise platform
Share the link to your md document with your reviewer
Replace
https://github.com/
For example, my username is ramnathv, so my link would be:
You can turn your plots into a beautiful dashboard with minimal effort. An Rmarkdown dashboard template has been provided for you. Open the file project-01-explore-babynames-flexdashboard.Rmd. It has placeholders for code, where you can paste the completed code chunks from this document. Pay close attention to the labels of the chunks referred to.