Alright, we've spent the last few dozen lessons building up your toolkit. Now, we're at the part I always dread but know is necessary: the timed mock assessment. Instead of just handing you a PDF of questions and saying "good luck," I want to walk you through how I approach these under pressure. I've put together a messy dataset of global coffee production—real-world style, which means it's formatted poorly—and I'm going to set a timer for 20 minutes to extract three specific insights.

Wrestling with the "Wide" Data Trap

First things first. I've got this `coffee_data` frame. Let's see what I'm working with.

# I'll start by peeking at the structure
head(coffee_data)
# Output:
# Country    2020  2021  2022  Region
# Brazil     3500  3200  3800  Americas
# Vietnam    1800  1900  1700  Asia
# Ethiopia    400   420   410  Africa

Immediately, I see the trap. The years are columns. If the assessment asks me to find the average growth rate across all years, I can't do that easily while the data is "wide." I'm tempted to just write a loop, but this is a Tidyverse assessment—I need to be faster. I'll pivot this immediately.

library(tidyverse)

# Attempt 1: Just pivot
coffee_long <- coffee_data %>% 
  pivot_longer(cols = 2:4, names_to = "year", values_to = "production")

head(coffee_long)
# Wait, the 'year' column is a character string now. 
# I can't do math on "2020".

That's a classic mistake. In a timed test, it's easy to forget that `pivot_longer` doesn't automatically know that your column names are actually numbers. I'll fix that on the fly using `mutate`.

coffee_long <- coffee_data %>% 
  pivot_longer(cols = 2:4, names_to = "year", values_to = "production") %>% 
  mutate(year = as.numeric(year))

The Joining Headache

Now, the mock prompt asks for the total production grouped by "Economic Tier" (High, Middle, Low), but that info isn't in my main table. It's in a separate `country_meta` dataframe. Let's try a simple join.

# Trying to merge the data
final_set <- coffee_long %>% 
  left_join(country_meta, by = "Country")

# Checking for NAs
sum(is.na(final_set$economic_tier))
# Output: 12

Twelve NAs? That's a red flag. I know every country in my production list should have a tier. I'll check the original `country_meta` and... there it is. Some countries are listed as "Viet Nam" in the meta-table but "Vietnam" in the production table. This happens constantly in real-world data. I don't have time to manually rename everything, so I'll use `str_replace` to standardize the most common culprit before re-joining.

country_meta <- country_meta %>% 
  mutate(Country = str_replace(Country, "Viet Nam", "Vietnam"))

# Now try the join again
final_set <- coffee_long %>% 
  left_join(country_meta, by = "Country")


Racing the Clock with Aggregation
The final requirement is a summary table showing the year-over-year growth percentage per region. I'm down to my last five minutes. I need to calculate the difference between the current year and the previous year. I'll use `lag()`, but I have to be careful—if I don't group by region first, I'll accidentally subtract Brazil's 2020 production from Ethiopia's 2021 production.

growth_report <- final_set %>% 
  group_by(Region) %>% 
  arrange(year) %>% 
  mutate(prev_prod = lag(production)) %>% 
  mutate(growth = (production - prev_prod) / prev_prod * 100) %>% 
  ungroup()

# Let's see if it worked
filter(growth_report, Region == "Americas")


There it is. The `lag()` function is a lifesaver here. If I had tried to do this with base R indexing, I'd probably still be staring at the screen. By chaining these Tidyverse verbs, I've turned a messy, wide dataset into a clean growth report in under 20 minutes. The key isn't knowing every function perfectly; it's knowing how to pivot, join, and group without getting tangled in the syntax.

📋 Practical Task

Exercise: Cleaning and Analyzing the Brazilian Coffee Export Surge

You have been provided with two dataframes: `export_data` (containing columns `Country`, `2015`, `2016`, `2017`, `2018`, `2019`, and `2020` representing export volumes) and `country_details` (containing `Country` and `Export_Tax_Bracket`).

Write a script to perform the following tasks:

• Transform the `export_data` from wide format to long format, ensuring the year is stored as an integer.
• Join the data with `country_details`, handling the case where "Brazil" is misspelled as "Brasil" in the `country_details` table.
• Calculate the total export volume for each `Export_Tax_Bracket` for the year 2020.
• Filter the final result to show only brackets that exported more than 1,000,000 units in 2020.