1 Challenge Description
This month, the challenge is to create and share a radial data visualization. Any circular view of data is welcome, but the data and scenario should lend itself well to this.
2 Dataset
Data came from Kaggle and includes historical hourly weather data for 36 different cities from 2012-2017. However, only Seattle data will be used in this project.
3 Setup
3.1 Load Libraries
if (!require("pacman")) install.packages("pacman")
pacman::p_load("tidyverse","visdat","lubridate","packcircles","ggforce")3.2 Import Data
df <- read_csv("weather_description.csv")4 Data Wrangling
- There are 45,253 observations.
- Each city has its own column with a character description of the weather pattern
- The first row is NA for all cities.
4.1 View Data
glimpse(df)## Observations: 45,253
## Variables: 37
## $ datetime <dttm> 2012-10-01 12:00:00, 2012-10-01 13:00:00,...
## $ Vancouver <chr> NA, "mist", "broken clouds", "broken cloud...
## $ Portland <chr> NA, "scattered clouds", "scattered clouds"...
## $ `San Francisco` <chr> NA, "light rain", "sky is clear", "sky is ...
## $ Seattle <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ `Los Angeles` <chr> NA, "mist", "sky is clear", "sky is clear"...
## $ `San Diego` <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ `Las Vegas` <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ Phoenix <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ Albuquerque <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ Denver <chr> NA, "light rain", "broken clouds", "broken...
## $ `San Antonio` <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ Dallas <chr> NA, "mist", "sky is clear", "sky is clear"...
## $ Houston <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ `Kansas City` <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ Minneapolis <chr> NA, "broken clouds", "broken clouds", "bro...
## $ `Saint Louis` <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ Chicago <chr> NA, "overcast clouds", "overcast clouds", ...
## $ Nashville <chr> NA, "mist", "overcast clouds", "overcast c...
## $ Indianapolis <chr> NA, "overcast clouds", "overcast clouds", ...
## $ Atlanta <chr> NA, "light rain", "overcast clouds", "over...
## $ Detroit <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ Jacksonville <chr> NA, "scattered clouds", "scattered clouds"...
## $ Charlotte <chr> NA, "mist", "overcast clouds", "overcast c...
## $ Miami <chr> NA, "light intensity drizzle", "broken clo...
## $ Pittsburgh <chr> NA, "mist", "scattered clouds", "scattered...
## $ Toronto <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ Philadelphia <chr> NA, "broken clouds", "broken clouds", "bro...
## $ `New York` <chr> NA, "few clouds", "few clouds", "few cloud...
## $ Montreal <chr> NA, "overcast clouds", "sky is clear", "sk...
## $ Boston <chr> NA, "sky is clear", "few clouds", "few clo...
## $ Beersheba <chr> NA, "sky is clear", "sky is clear", "overc...
## $ `Tel Aviv District` <chr> NA, "sky is clear", "sky is clear", "sky i...
## $ Eilat <chr> "haze", "haze", "broken clouds", "broken c...
## $ Haifa <chr> NA, "sky is clear", "overcast clouds", "ov...
## $ Nahariyya <chr> NA, "sky is clear", "sky is clear", "overc...
## $ Jerusalem <chr> NA, "sky is clear", "overcast clouds", "ov...head(df)## # A tibble: 6 x 37
## datetime Vancouver Portland `San Francisco` Seattle
## <dttm> <chr> <chr> <chr> <chr>
## 1 2012-10-01 12:00:00 <NA> <NA> <NA> <NA>
## 2 2012-10-01 13:00:00 mist scatter~ light rain sky is~
## 3 2012-10-01 14:00:00 broken c~ scatter~ sky is clear sky is~
## 4 2012-10-01 15:00:00 broken c~ scatter~ sky is clear sky is~
## 5 2012-10-01 16:00:00 broken c~ scatter~ sky is clear sky is~
## 6 2012-10-01 17:00:00 broken c~ scatter~ sky is clear sky is~
## # ... with 32 more variables: `Los Angeles` <chr>, `San Diego` <chr>, `Las
## # Vegas` <chr>, Phoenix <chr>, Albuquerque <chr>, Denver <chr>, `San
## # Antonio` <chr>, Dallas <chr>, Houston <chr>, `Kansas City` <chr>,
## # Minneapolis <chr>, `Saint Louis` <chr>, Chicago <chr>,
## # Nashville <chr>, Indianapolis <chr>, Atlanta <chr>, Detroit <chr>,
## # Jacksonville <chr>, Charlotte <chr>, Miami <chr>, Pittsburgh <chr>,
## # Toronto <chr>, Philadelphia <chr>, `New York` <chr>, Montreal <chr>,
## # Boston <chr>, Beersheba <chr>, `Tel Aviv District` <chr>, Eilat <chr>,
## # Haifa <chr>, Nahariyya <chr>, Jerusalem <chr>levels(as.factor(df$Seattle))## [1] "broken clouds" "drizzle"
## [3] "few clouds" "fog"
## [5] "haze" "heavy intensity drizzle"
## [7] "heavy intensity rain" "heavy intensity shower rain"
## [9] "heavy snow" "light intensity drizzle"
## [11] "light intensity shower rain" "light rain"
## [13] "light shower snow" "light snow"
## [15] "mist" "moderate rain"
## [17] "overcast clouds" "proximity thunderstorm"
## [19] "scattered clouds" "shower rain"
## [21] "sky is clear" "smoke"
## [23] "snow" "squalls"
## [25] "thunderstorm" "thunderstorm with heavy rain"
## [27] "thunderstorm with light rain" "thunderstorm with rain"
## [29] "very heavy rain"summary(df$datetime)## Min. 1st Qu. Median
## "2012-10-01 12:00:00" "2014-01-15 21:00:00" "2015-05-02 06:00:00"
## Mean 3rd Qu. Max.
## "2015-05-02 06:00:00" "2016-08-15 15:00:00" "2017-11-30 00:00:00"4.2 Filter Data
Data before December 2012 was removed in order to have an equal number of each month so that Winter will not show skewed data.
df <- df %>% select(datetime,Seattle)
df <- df %>% filter(datetime > "2012-11-30") 4.3 Missing Values
There are no missing values.
#Visualize missing values
vis_miss(df, sort_miss=TRUE)
4.4 Prepare Dataframe
df <- df %>% transmute(month = month(datetime),
season = case_when(month %in% c(12,1,2) ~ "Winter",
month %in% c(3,4,5) ~ "Spring",
month %in% c(6,7,8) ~ "Summer",
month %in% c(9,10,11) ~ "Fall"),
weather = case_when(str_detect(Seattle, "rain|drizzle") ~ "rain",
str_detect(Seattle, "mist|fog") ~ "fog",
str_detect(Seattle, "clear") ~ "clear",
str_detect(Seattle, "squalls") ~ "squalls",
str_detect(Seattle, "cloud") ~ "clouds",
str_detect(Seattle, "smoke|haze") ~ "haze",
str_detect(Seattle, "snow") ~ "snow",
str_detect(Seattle, "thunderstorm") ~ "thunder"))
# aster chart
df_aster <- df %>% group_by(weather) %>% summarize(count=n()) %>% arrange(count) %>% ungroup()
df_aster <- df_aster %>% mutate_at(vars(weather), as.factor)
df_aster$weather <- fct_reorder(df_aster$weather, df_aster$count)
df_aster <- df_aster %>% filter(count > 20)
# circle chart
df_circle <- df %>% group_by(season,weather) %>% summarize(count=n()) %>% arrange(count) %>% ungroup()
df_circle <- df_circle %>% mutate_if(is.character, as.factor)
df_circle <- df_circle %>% filter(count > 20)5 Visualizations
5.1 Aster Chart
df_aster %>% ggplot(aes(weather,sqrt(count),fill=weather))+
geom_col()+
geom_text(aes(y=50,label=weather))+
coord_polar()+
ylim(-10,125)+
labs(title="Seattle Weather Occurrences",subtitle="Measured from 2012 - 2017",fill="Weather Type")+
theme_minimal()+
theme(axis.title = element_blank(),
axis.text = element_blank(),
plot.title = element_text(hjust=0.5),
plot.subtitle = element_text(hjust=0.5),
legend.position = "none")
5.2 Circle Packed Chart
set.seed(333)
df_packs <- df_circle$count %>% circleRepelLayout() %>% .$layout
df_packs <- bind_cols(df_circle, df_packs) %>% select(-count)
# set colors
colors <- c("#BFD1E5","#8CC7A1","#C1666B","#D4B483") %>% set_names(nm = c("Winter","Spring","Summer","Fall"))
# main circle plot
p <- df_packs %>% ggplot() +
geom_circle(aes(x0=x, y0=y, r=radius, fill=season, color=season), size=0.1, show.legend=FALSE) +
geom_text(aes(x=x, y=y, label=weather, size=radius*0.5), show.legend=FALSE) +
scale_fill_manual(values=colors) +
scale_color_manual(values=colors) +
coord_fixed() +
ggtitle("Seattle Weather Occurrences")+
theme_void() +
theme(text = element_text(family="courier"),
plot.title = element_text(face="bold", hjust=0.5, size=14),
plot.background = element_rect(fill="snow", color="black"))
# make legend maually
legend <-
tibble(y=rep(1, each=4), x=rep(1:4, times=1), fill=colors %>% names(), r=0.1) %>%
ggplot() +
geom_text(aes(x=x, y=y-.2, label=fill), size=2.25, vjust=1, family="courier", fontface="bold") +
geom_circle(aes(x0=x, y0=y, fill=fill, color=fill, r=r), show.legend=FALSE) +
scale_fill_manual(values=colors) +
scale_colour_manual(values=colors) +
coord_fixed() +
theme_void() +
lims(x=c(.5,6), y=c(0.5, 2.3))
leg_text <- tibble(x=0.16, y=0.17, label="Season") %>%
ggplot()+
geom_text(aes(x=x, y=y, label=label), family="courier", fontface="bold", size=3.75, show.legend=FALSE)+
coord_fixed()+
lims(x=c(-4,5), y=c(-1, 1))+
theme_void()
# combine plot with legend
p + annotation_custom(
grob = ggplotGrob(legend),
xmin = -190,
xmax = -40,
ymin = -125,
ymax = -75) +
annotation_custom(
grob = ggplotGrob(leg_text),
xmin = -165,
xmax = -100,
ymin = -110,
ymax = -90)
# save final plot
ggsave("seattle-weather.png", width=6.1)