← Back to projects
R tidyverse ggplot2 EDA

Exploratory Data Analysis

Two open-ended R analyses: US regional minimum wage trends (1968–2020) and National Park trail exploration with functional programming and statistical inference.

Data

US minimum wage data from 1968–2020, with effective wages adjusted to 2020 dollars. A Region variable was added using US Census Bureau groupings: Northeast, Midwest, South, West.

library(tidyverse); library(ggplot2)
min_wage <- read_csv("minimum-wage-data.csv")
enriched_region_list <- read_csv("region_list1.txt", show_col_types = FALSE)
enriched_min_wage    <- full_join(min_wage, enriched_region_list)

Summary Statistics

2010–2020

enriched_min_wage |>
  filter(2010 < Year) |>
  group_by(Region) |>
  summarise(
    mean_effective_wage = mean(Effective.Minimum.Wage.2020.Dollars),
    var_effective_wage  = var(Effective.Minimum.Wage.2020.Dollars)) |>
  slice_max(var_effective_wage, n = 5)
Summary 2010-2020

West: highest mean ($9.09) and variance (2.04)

1968–1980

enriched_min_wage |>
  filter(1980 > Year) |>
  group_by(Region) |>
  summarise(
    mean_effective_wage = mean(Effective.Minimum.Wage.2020.Dollars),
    var_effective_wage  = var(Effective.Minimum.Wage.2020.Dollars)) |>
  slice_max(var_effective_wage, n = 5)
Summary 1968-1980

Visualizations

Average Minimum Wage by Region, 1968–2020

enriched_min_wage |>
  filter(Region != "Other") |>
  group_by(Region, Year) |>
  summarise(mean_effective_wage = mean(Effective.Minimum.Wage.2020.Dollars)) |>
  ggplot(aes(x = Year, y = mean_effective_wage, color = Region)) +
  geom_line() +
  labs(y = "Mean Minimum Wage", title = "Average Minimum Wage of US Regions 1968-2020")
Average Minimum Wage 1968-2020

Distribution by Region, 2010–2020

enriched_min_wage |>
  filter(Region != "Other", Year > 2009) |>
  ggplot(aes(x = factor(Year), y = Effective.Minimum.Wage.2020.Dollars, color = factor(Region))) +
  geom_boxplot(outlier.shape = NA) +
  guides(color = guide_legend(title = "Region"))
Minimum Wage Distribution 2010-2020

Analysis

From 1975 to 2000, regional minimum wages converged closely. Since 2000 they have diverged sharply, with the Northeast growing substantially while the South and Midwest lagged. From 2010–2020, year-over-year variance within each region grew steadily, with the West consistently showing the widest distribution, reflecting state-level policy divergence.

Data

All US National Park trails: state, area name, length, elevation gain, difficulty rating, average rating, popularity, visitor usage, and route type.

library(tidyverse); library(purrr)
national_park <- read.csv("national_park_trails.csv")

Exploration

Fourth longest trail in California?

national_park |>
  filter(state_name == "California") |>
  mutate(length_mi = length * 0.000621371) |>
  slice_max(n = 4, order_by = length_mi) |> slice(4) |> select(name, length_mi)
4th longest CA trail

Big SEKI Loop: 145.2 miles

Park with most "easy" trails? (elev < 500m, dist < 5km, difficulty < 4)

national_park |>
  mutate(length_km = length * 0.001) |>
  filter(elevation_gain < 500, difficulty_rating < 4, length_km < 5) |>
  group_by(area_name) |> summarise(easy_trail_num = n()) |>
  slice_max(n = 1, order_by = easy_trail_num)
Most easy trails

Parks with "Mountain" in name?

national_park |> filter(str_detect(area_name, "Mount|Mt|Mountain")) |> count()
Mountain count

Distribution of trail scores

national_park |>
  pivot_longer(cols = c(avg_rating, difficulty_rating, popularity, visitor_usage),
               names_to = "measure", values_to = "score") |>
  ggplot(aes(x = score)) + geom_density() + facet_wrap(~ measure, scales = 'free')
Trail score distributions

Trail Recommendation Function

recommend_trails <- function(data, park_name, min_elev=0, max_elev=Inf, min_rating=0, trl_type=NULL) {
  result <- data |>
    filter(area_name==park_name, elevation_gain>=min_elev,
           elevation_gain<=max_elev, avg_rating>=min_rating)
  if (!missing(trl_type)) result <- result |> filter(route_type %in% trl_type)
  result |> select(name)
}

# Test: Haleakala, min 1000m elev, min 4.0 rating
national_park |> recommend_trails("Haleakala National Park", min_elev=1000, min_rating=4)
Haleakala recommendations

map_dfr across 7 California parks

map_dfr(.x = c("Yosemite National Park", "Joshua Tree National Park",
               "Death Valley National Park", "Pinnacles National Park",
               "Lassen National Park", "Sequoia National Park", "King's Canyon National Park"),
        ~ recommend_trails(national_park, .x, min_elev=1000, max_elev=3000, min_rating=4.5, trl_type="loop"))
map_dfr results

Research Question: Yosemite vs. Yellowstone

Which park has "better" trails, defined by higher average rating and popularity among dog-friendly trails?

national_park |>
  filter(area_name %in% c("Yosemite National Park", "Yellowstone National Park"),
         avg_rating > 3, !str_detect(features, "'dogs-no'")) |>
  ggplot(aes(x=popularity, y=length, color=area_name)) + geom_line() +
  scale_color_manual(values=c("Yosemite National Park"="brown", "Yellowstone National Park"="#CD9600"))
Popularity vs Trail Length

Welch Two-Sample t-test

better_park <- national_park |>
  filter(area_name %in% c("Yosemite National Park", "Yellowstone National Park")) |>
  select(area_name, avg_rating, popularity)
t.test(avg_rating  ~ area_name, data=better_park)  # p > 0.05, not significant
t.test(popularity ~ area_name, data=better_park)  # p < 0.05, Yosemite higher
t-test rating

Rating: not significant (p > 0.05)

t-test popularity

Popularity: significant (p < 0.05)

Boxplot popularity

Yosemite's mean popularity (9.235) significantly exceeds Yellowstone's (7.001). Combined with greater trail variety, Yosemite is the better park by this definition.