Welcome to reshaping data in R using dplyr and tidyr. You’ll learn how to convert between wide and long formats, when to use each, and how to do calculations more efficiently with long-format data.
Before diving into the analysis, let’s load the necessary R packages. These packages will help us manipulate data efficiently.
Attaching package: 'dplyr'The following objects are masked from 'package:stats':
filter, lagThe following objects are masked from 'package:base':
intersect, setdiff, setequal, unionhere() starts at C:/IMF-R-BookThe same dataset can be written in two different formats: wide and long.
In a wide format, values do not repeat in the first column. This format works well if you have a time series of one variable for various countries, as each country’s data for different years can be spread across multiple columns.
Example of wide data:
wide_data <- data.frame(
year = c(2016, 2017, 2018, 2019),
consumption_BRA = c(5000, 5100, 5200, 5300),
consumption_MEX = c(4000, 4100, 4200, 4300),
consumption_USA = c(15000, 15500, 16000, 16500),
investment_BRA = c(1500, 1550, 1600, 1650),
investment_MEX = c(1200, 1250, 1300, 1350),
investment_USA = c(5000, 5100, 5200, 5300),
exports_BRA = c(1000, 1100, 1200, 1300),
exports_MEX = c(2000, 2100, 2200, 2300),
exports_USA = c(6000, 6200, 6400, 6600),
imports_BRA = c(800, 850, 900, 950),
imports_MEX = c(1000, 1050, 1100, 1150),
imports_USA = c(3000, 3100, 3200, 3300)
)
wide_data year consumption_BRA consumption_MEX consumption_USA investment_BRA
1 2016 5000 4000 15000 1500
2 2017 5100 4100 15500 1550
3 2018 5200 4200 16000 1600
4 2019 5300 4300 16500 1650
investment_MEX investment_USA exports_BRA exports_MEX exports_USA imports_BRA
1 1200 5000 1000 2000 6000 800
2 1250 5100 1100 2100 6200 850
3 1300 5200 1200 2200 6400 900
4 1350 5300 1300 2300 6600 950
imports_MEX imports_USA
1 1000 3000
2 1050 3100
3 1100 3200
4 1150 3300In a long format, values do repeat in the first column. This format is more efficient when you have multiple variables for different countries, as it allows for easier manipulation and analysis.
Example of long data:
long_data <- data.frame(
year = rep(2016:2019, each = 3),
country = rep(c("BRA", "MEX", "USA"), times = 4),
consumption = c(5000, 4000, 15000, 5100, 4100, 15500, 5200, 4200, 16000, 5300, 4300, 16500),
investment = c(1500, 1200, 5000, 1550, 1250, 5100, 1600, 1300, 5200, 1650, 1350, 5300),
exports = c(1000, 2000, 6000, 1100, 2100, 6200, 1200, 2200, 6400, 1300, 2300, 6600),
imports = c(800, 1000, 3000, 850, 1050, 3100, 900, 1100, 3200, 950, 1150, 3300)
)
long_data year country consumption investment exports imports
1 2016 BRA 5000 1500 1000 800
2 2016 MEX 4000 1200 2000 1000
3 2016 USA 15000 5000 6000 3000
4 2017 BRA 5100 1550 1100 850
5 2017 MEX 4100 1250 2100 1050
6 2017 USA 15500 5100 6200 3100
7 2018 BRA 5200 1600 1200 900
8 2018 MEX 4200 1300 2200 1100
9 2018 USA 16000 5200 6400 3200
10 2019 BRA 5300 1650 1300 950
11 2019 MEX 4300 1350 2300 1150
12 2019 USA 16500 5300 6600 3300Let’s start by reshaping a dataset from wide to long format. We’ll use the example dataset that contains GDP components (consumption, investment, exports, and imports) for multiple countries.
To reshape the wide data data into long format with columns year, country, consumption, investment, exports, and imports, we’ll use the pivot_longer() function from the tidyr package:
long_data <- wide_data %>%
pivot_longer(cols = -year,
names_to = c("variable", "country"),
names_sep = "_",
values_to = "value") %>%
pivot_wider(names_from = "variable", values_from = "value") %>%
arrange(country,year)
long_data# A tibble: 12 × 6
year country consumption investment exports imports
<dbl> <chr> <dbl> <dbl> <dbl> <dbl>
1 2016 BRA 5000 1500 1000 800
2 2017 BRA 5100 1550 1100 850
3 2018 BRA 5200 1600 1200 900
4 2019 BRA 5300 1650 1300 950
5 2016 MEX 4000 1200 2000 1000
6 2017 MEX 4100 1250 2100 1050
7 2018 MEX 4200 1300 2200 1100
8 2019 MEX 4300 1350 2300 1150
9 2016 USA 15000 5000 6000 3000
10 2017 USA 15500 5100 6200 3100
11 2018 USA 16000 5200 6400 3200
12 2019 USA 16500 5300 6600 3300If we want to reshape the long data back into wide format, we can use the pivot_wider() function:
wide_data_again <- long_data %>%
pivot_wider(names_from = "country",
values_from = c("consumption", "investment", "exports", "imports"))
wide_data_again# A tibble: 4 × 13
year consumption_BRA consumption_MEX consumption_USA investment_BRA
<dbl> <dbl> <dbl> <dbl> <dbl>
1 2016 5000 4000 15000 1500
2 2017 5100 4100 15500 1550
3 2018 5200 4200 16000 1600
4 2019 5300 4300 16500 1650
# ℹ 8 more variables: investment_MEX <dbl>, investment_USA <dbl>,
# exports_BRA <dbl>, exports_MEX <dbl>, exports_USA <dbl>, imports_BRA <dbl>,
# imports_MEX <dbl>, imports_USA <dbl>Let’s consider an example where we need to calculate GDP for multiple countries using the components (consumption, investment, exports, and imports).
In wide format, you would need to perform the calculation for each year and each country separately:
wide_data <- wide_data %>%
mutate(gdp_BRA = consumption_BRA + investment_BRA + exports_BRA - imports_BRA,
gdp_MEX = consumption_MEX + investment_MEX + exports_MEX - imports_MEX,
gdp_USA = consumption_USA + investment_USA + exports_USA - imports_USA)
wide_data year consumption_BRA consumption_MEX consumption_USA investment_BRA
1 2016 5000 4000 15000 1500
2 2017 5100 4100 15500 1550
3 2018 5200 4200 16000 1600
4 2019 5300 4300 16500 1650
investment_MEX investment_USA exports_BRA exports_MEX exports_USA imports_BRA
1 1200 5000 1000 2000 6000 800
2 1250 5100 1100 2100 6200 850
3 1300 5200 1200 2200 6400 900
4 1350 5300 1300 2300 6600 950
imports_MEX imports_USA gdp_BRA gdp_MEX gdp_USA
1 1000 3000 6700 6200 23000
2 1050 3100 6900 6400 23700
3 1100 3200 7100 6600 24400
4 1150 3300 7300 6800 25100This approach becomes cumbersome if you have many countries and years.
In long format, you can perform the calculation more efficiently:
# A tibble: 12 × 7
year country consumption investment exports imports gdp
<dbl> <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
1 2016 BRA 5000 1500 1000 800 6700
2 2017 BRA 5100 1550 1100 850 6900
3 2018 BRA 5200 1600 1200 900 7100
4 2019 BRA 5300 1650 1300 950 7300
5 2016 MEX 4000 1200 2000 1000 6200
6 2017 MEX 4100 1250 2100 1050 6400
7 2018 MEX 4200 1300 2200 1100 6600
8 2019 MEX 4300 1350 2300 1150 6800
9 2016 USA 15000 5000 6000 3000 23000
10 2017 USA 15500 5100 6200 3100 23700
11 2018 USA 16000 5200 6400 3200 24400
12 2019 USA 16500 5300 6600 3300 25100Using long format, the calculation is more straightforward and scalable, especially when dealing with large datasets.
You’ve learned how to reshape data between wide and long formats in R. Both formats have their uses, and long format simplifies complex calculations.