Texas housing data
- Subsetting and faceting
- Animated plots
Interplay of Environmental Factors and Ozone Levels

first_custom <- function() {
  theme_minimal() +
    theme(
      axis.line = element_line(color = "gray0"),
      plot.background  = element_rect(fill = "grey95"),
      panel.grid.major = element_line(linewidth = 0.5, color = "gray80"),
      panel.grid.minor = element_blank(),
      #panel.grid.minor = element_line(color = "gray85"),
      plot.margin = margin(15, 15, 15 ,15),
      plot.title = element_text(face = "bold"),
      plot.title.position = "plot",
      plot.caption = element_text(size = 8, face = "italic"),
      plot.caption.position = "plot",
      legend.background = element_rect(),
      legend.justification = (c(1,0)),
      legend.text = element_text(size = 8),
      legend.title = element_text(size = 8),
      legend.key.size = unit(0.25, "cm")
    )
}

Texas housing data

Information about the housing market in Texas

provided by the TAMU real estate center
in the `ggplot2 package

data(txhousing, package = "ggplot2")
unique(txhousing$city)

##  [1] "Abilene"               "Amarillo"              "Arlington"            
##  [4] "Austin"                "Bay Area"              "Beaumont"             
##  [7] "Brazoria County"       "Brownsville"           "Bryan-College Station"
## [10] "Collin County"         "Corpus Christi"        "Dallas"               
## [13] "Denton County"         "El Paso"               "Fort Bend"            
## [16] "Fort Worth"            "Galveston"             "Garland"              
## [19] "Harlingen"             "Houston"               "Irving"               
## [22] "Kerrville"             "Killeen-Fort Hood"     "Laredo"               
## [25] "Longview-Marshall"     "Lubbock"               "Lufkin"               
## [28] "McAllen"               "Midland"               "Montgomery County"    
## [31] "Nacogdoches"           "NE Tarrant County"     "Odessa"               
## [34] "Paris"                 "Port Arthur"           "San Angelo"           
## [37] "San Antonio"           "San Marcos"            "Sherman-Denison"      
## [40] "South Padre Island"    "Temple-Belton"         "Texarkana"            
## [43] "Tyler"                 "Victoria"              "Waco"                 
## [46] "Wichita Falls"

This dataset contains lots of information

we have monthly info on several different metrics,
in many cities,
over the course of 15 years

We can use this dataset to answer several different kinds of questions;

as scientists, our role is to figure out which parts of the dataset we need in order to answer certain questions,
and convey the story we want to tell/the story we see in the data

Three variables,

sales, volume, and median,
represent different quantities
but all convey a similar value
housing sales.

To begin, we’ll make a plot showing the evolution of sales over time.

this is a time series plot using a line chart
- process the time information into the correct format, so a time series can be plotted
show volume vs. time
show all the cities in the dataset (no color)
use color to highlight 4 different cities of interest
proper titling and labeling
- 4 cities of interest should be labeled, other cities don’t need to be

library(scales)

## 
## Attaching package: 'scales'

## The following object is masked from 'package:viridis':
## 
##     viridis_pal

## The following object is masked from 'package:purrr':
## 
##     discard

## The following object is masked from 'package:readr':
## 
##     col_factor

library(RColorBrewer)

txhousing$date <- as.Date(with(txhousing, paste(year, month, "01", sep="-")), "%Y-%m-%d")

cities_of_interest <- c("Houston", "El Paso", "Austin", "San Antonio")
colors <- brewer.pal(n = length(cities_of_interest), name = "Dark2")
names(colors) <- cities_of_interest

plot <- ggplot(txhousing, aes(x = date, y = volume / 1e9, group = city)) +
  geom_line(aes(color = city), size = 0.6) +
  scale_color_manual(values = c("grey", colors)) +
  scale_y_continuous(breaks = seq(0, 2.5, 0.5), labels = seq(0, 2.5, 0.5)) +
  scale_x_date(limits = c(min(txhousing$date), max(txhousing$date) + 800),
               expand = c(0.01, 0.01)) +
  labs(title = "Time Series of Housing Sales Volume in Texas Cities",
       x = "Year",
       y = "Sales Volume (USD billions)",
       color = "City") +
  guides(color = guide_legend(override.aes = list(size=2)))

plot <- plot + geom_line(data = filter(txhousing, city %in% cities_of_interest),
                         aes(color = city))

last_points <- txhousing %>%
  group_by(city) %>%
  filter(city %in% cities_of_interest) %>%
  summarize(volume = last(volume / 1e9), date = max(date))

plot + geom_text(data = last_points, aes(label = city, x = date, y = volume, color = city), 
                 size = 3, hjust = -0.1) +
  theme_minimal() +
  theme(legend.position = "none",
        panel.grid.major.x = element_blank(),
        panel.grid.minor.x = element_blank(),
        panel.grid.minor.y = element_blank(),
        axis.text.y = element_text(size = 8))

- The aesthetic choices include a distinct color scheme for key cities to stand out against a neutral grey background for others, moderate line sizing for clarity, and scaled y-axis labels for readability.

This is not a typical “spaghetti chart” because it strategically emphasizes only a few cities, preventing it from becoming too cluttered.
The visualization’s effectiveness is somewhat limited because while it highlights certain cities well, the remaining cities’ lines are indistinguishable from one another, potentially obscuring detailed comparisons among them.

Subsetting and faceting

Next, let’s look more closely at the 4 cities of interest that you selected.

create a faceted plot that shows each city on a different plot
show sales, volume, and median vs. time
use aesthetics wisely so that the information is clear and consistent across all plots

cities_data <- txhousing %>% 
  filter(city %in% c("Houston", "El Paso", "Austin", "San Antonio"))

cities_data <- cities_data %>%
  mutate(volume_scaled = volume / 1e6,
         median_scaled = median / 100)

max_value <- max(cities_data$sales, cities_data$volume_scaled, cities_data$median_scaled, na.rm = TRUE)

cities_long <- cities_data %>%
  select(city, date, sales, volume_scaled, median_scaled) %>%
  gather(key = "metric", value = "value", sales, volume_scaled, median_scaled)

plot3 <- ggplot(cities_long, aes(x = date, y = value, color = metric)) +
  geom_line() +
  facet_wrap(~ city, scales = "free_y") + 
  scale_color_viridis_d(direction = 1, option = "D", name = "Metric", labels = c("Median", "Sales", "Volume")) +
  labs(title = "Sales Metrics Over Time for Selected Texas Cities",
       subtitle = "Volume is scaled down by a factor of 1e6 and median sales price by 100.",
       x = "Year",
       y = "Values",
       color = "Metric") +
  theme_minimal() +
  theme(legend.position = "bottom",
        plot.title.position = "plot",
        axis.text = element_text(size = 8))

print(plot3)

plot2 <- ggplot(cities_long, aes(x = date, y = value, color = metric)) +
  geom_line() +
  facet_wrap(~ city) +
  scale_color_viridis_d(direction = 1, option = "D", name = "Metric", labels = c("Median", "Sales", "Volume")) +
  scale_y_continuous(trans = 'log10', labels = label_comma()) +
  labs(title = "Sales Metrics Over Time for Selected Texas Cities",
       subtitle = "Volume is scaled down by a factor of 1e6 and median sales price by 100.",
       x = "Year",
       y = "Log-Scaled Values",
       color = "Metric",
       caption = "Y-axis values are log-scaled.") +
  theme_minimal() +
  theme(legend.position = "bottom",
        plot.caption = element_text(face = "italic", size = 8),
        plot.title.position = "plot",
        axis.text = element_text(size = 8))

print(plot2)

Aesthetic choices for both plots:

Colors to distinguish metrics consistently across both plots. Log scale in one plot to manage wide-ranging values. Independent y-axes in the second plot to focus on individual city trends. Subtitles to inform about data transformations.

City choice:

No special reason; these cities were familiar to me and also provided diverse market insights.

Housing sales stats comparison across both plots:

Houston’s market is notably larger and more volatile. El Paso’s market is smaller with less fluctuation. Austin and San Antonio show growth, with Austin’s being more pronounced.

Conclusions from both plots:

The log scale plot shows relative trends and magnitudes clearly. The independent y-axes plot highlights individual city market behaviors. Each city’s housing market has unique characteristics and trends.

Animated plots

Next, let’s use animation to add a layer to our plots.

cities_long$year <- format(as.Date(cities_long$date), "%Y")
cities_long <- cities_long %>%
  mutate(year = as.Date(paste0(year, "-01-01")))

# Animated Plot 3
animated_plot3 <- ggplot(cities_long, aes(x = date, y = value, color = metric)) +
  geom_line(lwd = 1.1) +
  facet_wrap(~ city, scales = "free_y") + 
  scale_color_viridis_d(direction = 1, option = "D", name = "Metric", labels = c("Median", "Sales", "Volume")) +
  labs(title = "Sales Metrics Over Time for Selected Texas Cities",
       subtitle = "Volume is scaled down by a factor of 1e6 and median sales price by 100.",
       x = "Year",
       y = "Values",
       color = "Metric") +
  theme_minimal() +
  theme(legend.position = "bottom",
        plot.title.position = "plot",
        axis.text = element_text(size = 8)) +
  transition_reveal(along = date)

Sales Metrics

# Animated Plot 2
animated_plot2 <- ggplot(cities_long, aes(x = date, y = value, color = metric)) +
  geom_line(lwd = 1.1) +
  facet_wrap(~ city) +
  scale_color_viridis_d(direction = 1, option = "D", name = "Metric", labels = c("Median", "Sales", "Volume")) +
  scale_y_continuous(trans = 'log10', labels = scales::label_comma()) +
  labs(title = "Sales Metrics Over Time for Selected Texas Cities",
       subtitle = "Volume is scaled down by a factor of 1e6 and median sales price by 100.",
       x = "Year",
       y = "Log-Scaled Values",
       color = "Metric",
       caption = "Y-axis values are log-scaled.") +
  theme_minimal() +
  theme(legend.position = "bottom",
        plot.caption = element_text(face = "italic", size = 8),
        plot.title.position = "plot",
        axis.text = element_text(size = 8)) +
  transition_reveal(along = date)

Log Scaled

Animation adds a temporal dimension to the data, highlighting trends and changes over time in an engaging way. ________________________________________________________________________________

Interplay of Environmental Factors and Ozone Levels

Please Choose a dataset

this can be any dataset
the datasets package has some nice datasets

Make at least 3 different visualizations about the data

any ones you want
you will be graded on the story you tell and how well the visualizations are connected to make the story
you can arrange them together or separate
TELL YOUR STORY in your own words as you are creating the plots, guide us
You may use animations, please submit a .gif as well if you do

library(patchwork)
data(airquality)


p1 <- ggplot(airquality, aes(x = Temp, y = Ozone)) +
  geom_point(alpha = 0.5) +
  geom_smooth(method = "lm", se = FALSE, color = "red") +
  labs(title = "Ozone vs. Temperature", x = "Temperature (degrees F)") +
  theme(title = element_text(size = 10))

# Plot 2: Ozone vs. Wind Speed
p2 <- ggplot(airquality, aes(x = Wind, y = Ozone)) +
  geom_point(alpha = 0.5) +
  geom_smooth(method = "lm", se = FALSE, color = "#35b779") +
  labs(title = "Ozone vs. Wind Speed", x = "Wind (mph)") +
  theme(title = element_text(size = 10))


p3 <- ggplot(airquality, aes(x = factor(Month), y = Ozone)) +
  geom_boxplot() +
  labs(title = "Monthly Ozone Levels", x = "Month") +
  theme(title = element_text(size = 10))

p4 <- ggplot(airquality, aes(x = Solar.R, y = Ozone)) +
  geom_point(alpha = 0.5) +
  geom_smooth(method = "lm", se = FALSE, color = "#fde725") +
  labs(title = "Ozone vs. Solar Radiation", x = "Solar Radiation (Langley)") +
  theme(title = element_text(size = 10))


combined_plots <- (p1 | p2) / (p3 | p4) +
  plot_annotation(
    title = "Interplay of Environmental Factors and Ozone Levels",
    subtitle = "Temperature and solar radiation elevate ozone levels, while wind reduces them.",
    theme = theme(plot.title = element_text(size = 14, face = "bold"), plot.subtitle = element_text(size = 10))
  )

combined_plots

## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'
## `geom_smooth()` using formula = 'y ~ x'

airquality <- mutate(airquality, Date = make_date(1973, Month, Day))

# Remove rows with NA values
airquality <- na.omit(airquality)

# Convert dataset to long format
airquality_long <- airquality %>%
  select(Date, Ozone, Solar.R, Wind, Temp) %>%
  pivot_longer(cols = -Date, names_to = "variable", values_to = "value")

# Rename variables with units
names_with_units <- c(Ozone = "Ozone (ppb)", Solar.R = "Solar Radiation (lang)",
                      Wind = "Wind (mph)", Temp = "Temperature (degrees F)")

# Apply labels
airquality_long$variable <- factor(airquality_long$variable,
                                   levels = names(names_with_units),
                                   labels = names_with_units)

# Define colors
colors <- c("Ozone (ppb)" = "#440154", "Solar Radiation (lang)" = "#fde725",
            "Wind (mph)" = "#35b779", "Temperature (degrees F)" = "red")

# Create the animated plot
animated_plot <- ggplot(airquality_long, aes(x = Date, y = value, color = variable)) +
  geom_line(lwd = 1.1) +
  scale_color_manual(values = colors) + 
  facet_wrap(~ variable, scales = "free_y", ncol = 1) + 
  labs(title = "Environmental Metrics Over Time",
       x = "Date",
       y = "Values") +
  theme_minimal() +
  theme(legend.position = "none", 
        axis.text = element_text(size = 8),
        strip.text.x = element_text(size = 10)) + 
  transition_reveal(along = Date)

Our story about air quality shows how the environment affects ozone levels through the following plot points:

“Ozone vs. Temperature” shows higher temperatures boosting ozone.
“Ozone vs. Wind Speed” reveals that wind disperses ozone, cleaning the air.
“Monthly Ozone Levels” display a clear seasonal spike in ozone during warmer months.
“Ozone vs. Solar Radiation” connects bright sunlight to increased ozone.

An animated plot, “Environmental Metrics Over Time,” brings motion to the story, dynamically showcasing how these elements fluctuate day-to-day and month-to-month.

This collection of images and animations essentially narrates the succinct tale of how temperature, wind, and sun interact with ozone throughout the year.

Multiple-Approaches

Ibrahim Uruc Tarim

07 June, 2024

Texas housing data

Subsetting and faceting

Animated plots

Interplay of Environmental Factors and Ozone Levels