5. Data Visualization (Line Graph)
Masahiko Asano
2021-09-13
Load the following packages
library(DT)
library(gapminder)
library(gghighlight)
library(stargazer)
library(tidyverse)Line Graph
- x-axis : Continuous variable, Date, etc
- The value of x-axis must be unique; no overlap
- y-axis : Continuous variable
We are going to deal with the following four cases:
- Averaged life Expectancy between Japanese and Chinese
- Median Age of LDP and DPJ Winners
- Political Polarization in US Congress
- Data on COVID-19
1. Life Expectancy (Japan & China)
1.1 Data (gapminder)
・gapminder is a built-in data set in R
(1) year |
: 1952-2007 (every 5 years) |
(2) lifeExp |
: Life expectancy |
(3) country |
: Country name |
(4) continent |
: Continent name |
(5) pop |
: Population |
(6) gdpPercap |
: GDP per capita |
- Loard
gapminder
library(gapminder)- Look into the gapminder by using
glimpsefunction
data(gapminder)
glimpse(gapminder)Rows: 1,704
Columns: 6
$ country <fct> "Afghanistan", "Afghanistan", "Afghanistan", "Afghanistan", …
$ continent <fct> Asia, Asia, Asia, Asia, Asia, Asia, Asia, Asia, Asia, Asia, …
$ year <int> 1952, 1957, 1962, 1967, 1972, 1977, 1982, 1987, 1992, 1997, …
$ lifeExp <dbl> 28.801, 30.332, 31.997, 34.020, 36.088, 38.438, 39.854, 40.8…
$ pop <int> 8425333, 9240934, 10267083, 11537966, 13079460, 14880372, 12…
$ gdpPercap <dbl> 779.4453, 820.8530, 853.1007, 836.1971, 739.9811, 786.1134, …・Show the summary of gapminder
summary(gapminder) country continent year lifeExp
Afghanistan: 12 Africa :624 Min. :1952 Min. :23.60
Albania : 12 Americas:300 1st Qu.:1966 1st Qu.:48.20
Algeria : 12 Asia :396 Median :1980 Median :60.71
Angola : 12 Europe :360 Mean :1980 Mean :59.47
Argentina : 12 Oceania : 24 3rd Qu.:1993 3rd Qu.:70.85
Australia : 12 Max. :2007 Max. :82.60
(Other) :1632
pop gdpPercap
Min. :6.001e+04 Min. : 241.2
1st Qu.:2.794e+06 1st Qu.: 1202.1
Median :7.024e+06 Median : 3531.8
Mean :2.960e+07 Mean : 7215.3
3rd Qu.:1.959e+07 3rd Qu.: 9325.5
Max. :1.319e+09 Max. :113523.1
・Show the list of country
unique(gapminder$country) [1] Afghanistan Albania Algeria
[4] Angola Argentina Australia
[7] Austria Bahrain Bangladesh
[10] Belgium Benin Bolivia
[13] Bosnia and Herzegovina Botswana Brazil
[16] Bulgaria Burkina Faso Burundi
[19] Cambodia Cameroon Canada
[22] Central African Republic Chad Chile
[25] China Colombia Comoros
[28] Congo, Dem. Rep. Congo, Rep. Costa Rica
[31] Cote d'Ivoire Croatia Cuba
[34] Czech Republic Denmark Djibouti
[37] Dominican Republic Ecuador Egypt
[40] El Salvador Equatorial Guinea Eritrea
[43] Ethiopia Finland France
[46] Gabon Gambia Germany
[49] Ghana Greece Guatemala
[52] Guinea Guinea-Bissau Haiti
[55] Honduras Hong Kong, China Hungary
[58] Iceland India Indonesia
[61] Iran Iraq Ireland
[64] Israel Italy Jamaica
[67] Japan Jordan Kenya
[70] Korea, Dem. Rep. Korea, Rep. Kuwait
[73] Lebanon Lesotho Liberia
[76] Libya Madagascar Malawi
[79] Malaysia Mali Mauritania
[82] Mauritius Mexico Mongolia
[85] Montenegro Morocco Mozambique
[88] Myanmar Namibia Nepal
[91] Netherlands New Zealand Nicaragua
[94] Niger Nigeria Norway
[97] Oman Pakistan Panama
[100] Paraguay Peru Philippines
[103] Poland Portugal Puerto Rico
[106] Reunion Romania Rwanda
[109] Sao Tome and Principe Saudi Arabia Senegal
[112] Serbia Sierra Leone Singapore
[115] Slovak Republic Slovenia Somalia
[118] South Africa Spain Sri Lanka
[121] Sudan Swaziland Sweden
[124] Switzerland Syria Taiwan
[127] Tanzania Thailand Togo
[130] Trinidad and Tobago Tunisia Turkey
[133] Uganda United Kingdom United States
[136] Uruguay Venezuela Vietnam
[139] West Bank and Gaza Yemen, Rep. Zambia
[142] Zimbabwe
142 Levels: Afghanistan Albania Algeria Angola Argentina Australia ... Zimbabwe- Let’s say, we want to draw the graph like this
- Averaged life expectancy in Japan and China since 1950 to date
・We need the following three variables to draw the graph
(1) year |
: 1952-2007 (every 5 years) |
(2) lifeExp |
: Life expectancy |
(3) country |
: Country name |
1.2 Life Expectancy Data for Japanese
・Select the data for Japanese and name the data frame Japan
Japan <- gapminder %>%
dplyr::filter(country == "Japan") %>%
dplyr::select(year, lifeExp)
Japan # A tibble: 12 x 2
year lifeExp
<int> <dbl>
1 1952 63.0
2 1957 65.5
3 1962 68.7
4 1967 71.4
5 1972 73.4
6 1977 75.4
7 1982 77.1
8 1987 78.7
9 1992 79.4
10 1997 80.7
11 2002 82
12 2007 82.6・Draw the graph for Japanese life expectancy
ggplot(Japan, aes(x = year, y = lifeExp)) +
geom_point() +
geom_line()
1.3 Life Expectancy (Japan & China)
・Select the data for Japanese and Chinese and name the data frame jpn.chi
jpn.chi <- gapminder %>%
dplyr::filter(country == "China" | country == "Japan") %>%
dplyr::select(year, country, lifeExp)
jpn.chi# A tibble: 24 x 3
year country lifeExp
<int> <fct> <dbl>
1 1952 China 44
2 1957 China 50.5
3 1962 China 44.5
4 1967 China 58.4
5 1972 China 63.1
6 1977 China 64.0
7 1982 China 65.5
8 1987 China 67.3
9 1992 China 68.7
10 1997 China 70.4
# … with 14 more rows- Check the data frame
jpn.chi
DT::datatable(jpn.chi)- Draw a line graph including both Japanese and Chinese averaged life expectancy
ggplot(jpn.chi, aes(x = year, y = lifeExp, color = country)) +
geom_point() +
geom_line() +
labs(x = "Year", y = "Life Expectanct",
title = "Averaged Life Expectanct (1952-2007): Japanese and Chinese") 
1.4 Exercise
Question: Use the bulit-in data setgapminder, select the two countries you like and draw a line graph showing life expectancy for the two countries.
2. Winner’s Median Age
2.1 Data (hr96-17.csv)
- Download the Japanese Lower House Election Data hr96-17.csv and read on RStudio
- Make a new folder, named
datawithin your RProjct folder
- Put the
hr96-17.csvintodata
- You need to use
tidyversepackage to read csv.file
→ Loadtidyverse
library(tidyverse)- Read the csv file and name it
hr
hr <- read_csv("data/hr96-17.csv",
na = ".") # replace missing data with "." - Show the list of variables included in
hr
names(hr) [1] "year" "pref" "ku" "kun"
[5] "mag" "rank" "wl" "nocand"
[9] "seito" "j_name" "name" "term"
[13] "gender" "age" "exp" "status"
[17] "vote" "voteshare" "eligible" "turnout"
[21] "castvotes" "seshu_dummy" "jiban_seshu" "nojiban_seshu"| variable | detail |
|---|---|
| year | Election year (1996-2017) |
| age | Candidate’s age |
| wl | 0 = loser / 1 = single-member district (smd) winner / 2 = zombie winner |
| seito | Candidate’s affiliated party |
- Let’s say, we want to draw the graph like this
- Median Age of LDP and DPJ Winners since 1990 to date
- To draw the graph above, we need a variable,
age.median
age.median: Winner’s median age for the LDP and DPJ by electionsが必要
(1) year |
: Election year (1996-2017) |
(2) age.median |
: Candidate’s median age |
| (3) wl | 0 = loser / 1 = single-member district (smd) winner / 2 = zombie winner |
| (4) seito | Candidate’s affiliated party |
age.medianis not included in the date framehr
=> We need to make it
- Calculate the median of age per election and per party
- The data frame we need looks like this
| # | year | age.median | seito |
| 1 | 1996 | numeric | LDP(自民) |
| ・ | ・ | ・ | ・ |
| ・ | ・ | ・ | ・ |
| 7 | 2017 | numeric | LDP(自民) |
| 8 | 1996 | numeric | DPJ(民主) |
| ・ | ・ | ・ | ・ |
| ・ | ・ | ・ | ・ |
| 14 | 2014 | numeric | DPJ(民主) |
・Note: the DPJ was dissolved just prior to the 2017 lower house election
2.2 Make a variable age.median
- Calculate the median age of candidates per election (1996-2017)
Using dplyr package
・Check the variable, age prior to calculation
summary(hr$age) Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
25.0 43.0 51.0 50.9 59.0 94.0 4 seito.median <- hr %>% # Save as seito.median
dplyr::filter(seito == "自民" | seito == "民主") %>% # 自民党と民主党だけを残す
dplyr::filter(wl == 1) %>% # we only need winners(wl = 1)
drop_na(age) %>% # Delete the 5 missing valus (NA's)
group_by(year, seito) %>% # Calculate the median by year & by seito
summarise(age.median = median(age)) - Check the data
DT::datatable(seito.median)・We got the data needed to draw the line graph
2.3 Draw a Line Graph
ggplot(seito.median, aes(x = year, y = age.median,
color = seito, linetype = seito, shape = seito)) +
geom_point() +
geom_line() +
lims(y = c(40, 60)) + # y 軸の範囲を指定
ggtitle("Winner's Median Age by Party: 1996-2017") +
scale_color_discrete(name ="Party",
labels = c("DPJ", "LDP")) +
scale_linetype_discrete(name ="Party",
labels = c("DPJ", "LDP")) +
scale_shape_discrete(name ="Party",
labels = c("DPJ", "LDP")) +
ggtitle("Winner's Median Age by Party: 1996-2017") +
labs(x = "Election Year", y = "Median Age") +
theme_bw(base_family = "HiraKakuProN-W3")
2.4 Exercise
Question 1: Using the Japanese election data hr96-17.csv), draw a line graph of Winner’s voteshare (median) between 1996 and 2017 for the LDP and the DPJ.
Question 2: What could you say about the line graph?
(1) year |
: Election year (1996-2017) |
| (2) voteshare | Voteshare (%) |
| (3) wl | 0 = loser / 1 = single-member district (smd) winner / 2 = zombie winner |
| (4) seito | Candidate’s affiliated party |
3. Polarization in the US Congress
3.1 Data (congress.csv)
- The 80the Congress (1947-1948) - The 112nd Congress (2011-2012)
- Draw politician’s most ideal points on policy
DW-NOMINATE score
dwnom1 |
: Economic Issue ・・・ -1 (liberal) 〜 1 (consevative) |
dwnom2 |
: Race Issue ・・・ -1 (liberal) 〜 1 (conservative) |
- Sample size: 14552
Source: Nolan McCarty, Keith T. Poole, and Howard Rosenthal (2006) Polarized America:The Dance of Ideology and Unequal Riches. MIT Press.
Using Economic dimention (dwnom1), draw a graph of median by party
- Let’s say we want to draw a graph like this
3.2 Make a variable econ.median
- We need the following three variables to draw the line graph
(1) congress |
: Congressional session number (80-112) |
(2) econ.median |
: Politician’s most ideal median points on economy |
(3) party |
: Party of the congressional representative |
econ.medianis not included in the date frame
=> We need to make it
- Calculate the median of Politician’s most ideal median points on economy per congressional session
- The data frame we need looks like this
- Download the data congress.csv)
- Make a folder within your R Project folder and name it
data
- Put the
congress.csvin thedatafolder
- To read the
congress.csv, we need to loadreadrpackage, which is included intidyversepackage
library(tidyverse)- Read the data and name it
US
US <- read_csv("data/congress.csv")- Check the data before calculation
summary(US$congress) Min. 1st Qu. Median Mean 3rd Qu. Max.
80.00 88.00 96.00 96.01 104.00 112.00 - No missing data (
NA)
US <- US %>%
dplyr::filter(party == "Republican" | party == "Democrat") %>%
group_by(congress, party) %>%
summarise(econ.median = median(dwnom1)) - Check the dataframe
DT::datatable(US)Draw a line graph
ggplot(US, aes(x = congress, y = econ.median, color = party)) +
geom_point() +
geom_line() +
theme_bw() +
ggtitle("Political Polarization: ECON Dimention(US Congress:1947-2012)") +
labs(x = "Congress",
y = "DW-NOMINATE score (economic dimention)") 
Conclusion ・We can see how the economy score changes over time
・ We can see a clear polarization between two parties
・ The political polarization started around the 95th Congress session
・ Recently, the Democrat becomes more liberal (-) and the Republican becomes more conservative (+)
3.3 Exercise
Question 1: Using the variable dwnom2, draw a line graph to check whether we can see any Race polarization in the US Congress (the 80th - the 112nd) between the Democrat and the Republican.
Question 2: What could you say about the US polarization in Race issue?
Source: Nolan McCarty, Keith T. Poole, and Howard Rosenthal (2006) Polarized America:The Dance of Ideology and Unequal Riches. MIT Press.
- Data:
DW-NOMINATE score
- Download congress.csv)
dwnom1 |
: Economic Issue ・・・ -1 (liberal) 〜 1 (consevative) |
dwnom2 |
: Race Issue ・・・ -1 (liberal) 〜 1 (conservative) |
- Sample size: 14552
4. COVID-19
4.1 Data (COVID19_Worldwide.csv)
- We use the data Dr. Song made COVID19_Worldwide.csv
| Variables | Details |
|---|---|
| ID | ID |
| Country | Country name |
| Date | Date survey conducted |
| Confirmed_Day | The number of people infected COVID-19 per day |
| Confirmed_Total | The cummulated number of people infected COVID-19 (total) |
| Death_Day | The number of death of COVID-19 per day |
| Death_Total | The cummulated number of people died of COVID-19 (total) |
| Test_Day | The number of people newly tested COVID-19 per day |
| Test_Total | The cummulated number of people tested COVID-19 (total) |
COVID19_df <- read_csv("data/COVID19_Worldwide.csv",
guess_max = 10000)
# Set to judge the class after reading the first 10000 cases- Check the class of variables
str(COVID19_df)spec_tbl_df [31,806 × 9] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
$ ID : num [1:31806] 1 2 3 4 5 6 7 8 9 10 ...
$ Country : chr [1:31806] "Afghanistan" "Afghanistan" "Afghanistan" "Afghanistan" ...
$ Date : chr [1:31806] "2020/1/22" "2020/1/23" "2020/1/24" "2020/1/25" ...
$ Confirmed_Day : num [1:31806] 0 0 0 0 0 0 0 0 0 0 ...
$ Confirmed_Total: num [1:31806] 0 0 0 0 0 0 0 0 0 0 ...
$ Death_Day : num [1:31806] 0 0 0 0 0 0 0 0 0 0 ...
$ Death_Total : num [1:31806] 0 0 0 0 0 0 0 0 0 0 ...
$ Test_Day : num [1:31806] NA NA NA NA NA NA NA NA NA NA ...
$ Test_Total : num [1:31806] NA NA NA NA NA NA NA NA NA NA ...
- attr(*, "spec")=
.. cols(
.. ID = col_double(),
.. Country = col_character(),
.. Date = col_character(),
.. Confirmed_Day = col_double(),
.. Confirmed_Total = col_double(),
.. Death_Day = col_double(),
.. Death_Total = col_double(),
.. Test_Day = col_double(),
.. Test_Total = col_double()
.. )- Since the class of Date is
chr, we need to change it intoDate
COVID19_df <- COVID19_df %>%
mutate(Date = as.Date(Date))- Show the descriptive statistics of COVID19_df
library(stargazer)- Type
{r, results = "asis"}as chunk option
stargazer(as.data.frame(COVID19_df),
type ="html",
digits = 2)| Statistic | N | Mean | St. Dev. | Min | Pctl(25) | Pctl(75) | Max |
| ID | 31,806 | 16,124.51 | 9,379.38 | 1 | 7,952.2 | 24,235.8 | 32,413 |
| Confirmed_Day | 31,806 | 392.96 | 2,430.30 | -10,034 | 0 | 58 | 66,627 |
| Confirmed_Total | 31,806 | 18,250.14 | 115,471.60 | 0 | 0 | 2,505.8 | 3,184,582 |
| Death_Day | 31,806 | 17.61 | 112.88 | -1,918 | 0 | 1 | 2,614 |
| Death_Total | 31,806 | 1,039.01 | 6,565.51 | 0 | 0 | 55 | 134,094 |
| Test_Day | 8,281 | 14,332.71 | 61,911.94 | -3,743.00 | 839.00 | 9,041.00 | 2,022,722.00 |
| Test_Total | 8,686 | 559,831.30 | 2,146,252.00 | 1.00 | 21,545.25 | 353,916.50 | 39,011,749.00 |
DT::datatable(COVID19_df)4.2 Draw a Line Graph
- Make a plot with the raw data
- x-axis —
Date
- y-axis — cummulated death （
Death_Total）
- Take the logarithm of
Death_Total
=> The plot gets easier to see
COVID19_df %>%
ggplot() +
geom_line(aes(x = Date,
y = Death_Total)) +
scale_y_continuous(breaks = c(10, 100, 1000, 10000, 100000, 1000000),
labels = c("10", "100", "1000", "10000",
"100000", "1000000"),
trans = "log10") +
labs(x = "Month", y = "Cummurated number death", color = "Country") 
- This is not easy to see because
Date(x-axis) has overlappping values
- There are too many countries
- We need to choose small number of countries interested
- Show the list of countries
unique(COVID19_df$Country) [1] "Afghanistan" "Albania"
[3] "Algeria" "Andorra"
[5] "Angola" "Antigua and Barbuda"
[7] "Argentina" "Armenia"
[9] "Australia" "Austria"
[11] "Azerbaijan" "Bahamas"
[13] "Bahrain" "Bangladesh"
[15] "Barbados" "Belarus"
[17] "Belgium" "Belize"
[19] "Benin" "Bhutan"
[21] "Bolivia" "Bosnia and Herzegovina"
[23] "Botswana" "Brazil"
[25] "Brunei" "Bulgaria"
[27] "Burkina Faso" "Burma"
[29] "Burundi" "Cabo Verde"
[31] "Cambodia" "Cameroon"
[33] "Canada" "Central African Republic"
[35] "Chad" "Chile"
[37] "China" "Colombia"
[39] "Comoros" "Congo (Brazzaville)"
[41] "Congo (Kinshasa)" "Costa Rica"
[43] "Cote d'Ivoire" "Croatia"
[45] "Cuba" "Cyprus"
[47] "Czechia" "Denmark"
[49] "Djibouti" "Dominica"
[51] "Dominican Republic" "Ecuador"
[53] "Egypt" "El Salvador"
[55] "Equatorial Guinea" "Eritrea"
[57] "Estonia" "Eswatini"
[59] "Ethiopia" "Fiji"
[61] "Finland" "France"
[63] "Gabon" "Gambia"
[65] "Georgia" "Germany"
[67] "Ghana" "Greece"
[69] "Grenada" "Guatemala"
[71] "Guinea" "Guinea-Bissau"
[73] "Guyana" "Haiti"
[75] "Holy See" "Honduras"
[77] "Hungary" "Iceland"
[79] "India" "Indonesia"
[81] "Iran" "Iraq"
[83] "Ireland" "Israel"
[85] "Italy" "Jamaica"
[87] "Japan" "Jordan"
[89] "Kazakhstan" "Kenya"
[91] "South Korea" "Kosovo"
[93] "Kuwait" "Kyrgyzstan"
[95] "Laos" "Latvia"
[97] "Lebanon" "Lesotho"
[99] "Liberia" "Libya"
[101] "Liechtenstein" "Lithuania"
[103] "Luxembourg" "Madagascar"
[105] "Malawi" "Malaysia"
[107] "Maldives" "Mali"
[109] "Malta" "Mauritania"
[111] "Mauritius" "Mexico"
[113] "Moldova" "Monaco"
[115] "Mongolia" "Montenegro"
[117] "Morocco" "Mozambique"
[119] "Namibia" "Nepal"
[121] "Netherlands" "New Zealand"
[123] "Nicaragua" "Niger"
[125] "Nigeria" "North Macedonia"
[127] "Norway" "Oman"
[129] "Pakistan" "Panama"
[131] "Papua New Guinea" "Paraguay"
[133] "Peru" "Philippines"
[135] "Poland" "Portugal"
[137] "Qatar" "Romania"
[139] "Russia" "Rwanda"
[141] "Saint Kitts and Nevis" "Saint Lucia"
[143] "Saint Vincent and the Grenadines" "San Marino"
[145] "Sao Tome and Principe" "Saudi Arabia"
[147] "Senegal" "Serbia"
[149] "Seychelles" "Sierra Leone"
[151] "Singapore" "Slovakia"
[153] "Slovenia" "Somalia"
[155] "South Africa" "South Sudan"
[157] "Spain" "Sri Lanka"
[159] "Sudan" "Suriname"
[161] "Sweden" "Switzerland"
[163] "Syria" "Taiwan"
[165] "Tajikistan" "Tanzania"
[167] "Thailand" "Timor-Leste"
[169] "Togo" "Trinidad and Tobago"
[171] "Tunisia" "Turkey"
[173] "Uganda" "Ukraine"
[175] "United Arab Emirates" "United Kingdom"
[177] "United States" "Uruguay"
[179] "Uzbekistan" "Venezuela"
[181] "Vietnam" "West Bank and Gaza"
[183] "Western Sahara" "Yemen"
[185] "Zambia" "Zimbabwe" 4.3 Plot with small number of countries
G7 <- c("Cananda", "France", "Germany", "Italy", "Japan",
"United Kingdom", "United States")
COVID19_df %>%
filter(Country %in% G7) %>%
ggplot() +
geom_line(aes(x = Date,
y = Death_Total,
color = Country)) +
scale_y_continuous(breaks = c(10, 100, 1000, 10000, 100000, 1000000),
labels = c("10", "100", "1000", "10000",
"100000", "1000000"),
trans = "log10") +
labs(x = "Month", y = "Cummurated number death", color = "Country") 
- It gets way easier to see than the previous line graph
- But, the legend does not match the order of lines
- Fix it using
fct_reorder2()function
- Let’s fix it
COVID19_df %>%
filter(Country %in% G7) %>%
mutate(Country = fct_reorder2(Country,
Date,
Death_Total,
last2)) %>%
ggplot() +
geom_line(aes(x = Date,
y = Death_Total,
color = Country)) +
scale_y_continuous(breaks = c(10, 100, 1000, 10000, 100000, 1000000),
labels = c("10", "100", "1000", "10000",
"100000", "1000000"),
trans = "log10") +
labs(x = "Month", y = "Cummurated number death", color = "Country") 
- It gets much better!
4.4 Highlight particular countries
- Using
gglighlightpackage, you can put emphasis on certain coutries
- Let’s pick up 5 countries and draw a line graph
library(gghighlight)COVID19_df %>%
ggplot() +
geom_line(aes(x = Date,
y = Death_Total,
color = Country)) +
gghighlight(Country %in% c("Japan", "China", "South Korea",
"United States", "Taiwan")) +
scale_y_continuous(breaks = c(10, 100, 1000, 10000, 100000, 1000000),
labels = c("10", "100", "1000", "10000",
"100000", "1000000"),
trans = "log10") +
labs(x = "Month", y = "Cummurated number death", color = "Country") 
4.5 Exercise
Question 1: Pick up 5 countries you like and draw a line graph of cummurated number of death and date, highliting them.
Question 2: What could you say about the ling graph you get?
- 宋財泫 (Jaehyun Song)・矢内勇生 (Yuki Yanai)「私たちのR: ベストプラクティスの探究」
- 土井翔平（北海道大学公共政策大学院）「Rで計量政治学入門」
- 矢内勇生（高知工科大学）授業一覧
- 浅野正彦, 矢内勇生.『Rによる計量政治学』オーム社、2018年
- 浅野正彦, 中村公亮.『初めてのRStudio』オーム社、2018年
- Winston Chang, R Graphics Cookbook, O’Reilly Media, 2012.
- Kieran Healy, DATA VISUALIZATION, Princeton, 2019
- Kosuke Imai, Quantitative Social Science: An Introduction, Princeton University Press, 2017