Interactive plots with plotly

Building Dashboards with shinydashboard

Png Kee Seng

Researcher

An example: Airbnb listings in London

Imagine we are working at a tour agency based in London

A collaboration was set up with Airbnb

library(tidyverse)
listings <- read_csv("../data/listings.csv")

...
 $ host_name                     : chr [1:69351] "Susie" "Alina" "Luca" "Simon" ...
 $ neighbourhood_group           : logi [1:69351] NA NA NA NA NA NA ...
 $ neighbourhood                 : chr [1:69351] "Tower Hamlets" "Islington" "Tower Hamlets" "Islington" ...
 $ latitude                      : num [1:69351] 51.5 51.6 51.5 51.5 51.5 ...
 $ longitude                     : num [1:69351] -0.054 -0.1127 -0.0743 -0.1048 -0.2 ...
 $ room_type                     : chr [1:69351] "Private room" "Private room" "Entire home/apt" "Private room" ...
 $ price                         : num [1:69351] 55 50 90 180 297 75 204 379 90 30 ...
 $ minimum_nights                : num [1:69351] 3 1 5 4 14 3 2 4 3 7 ...
 $ number_of_reviews             : num [1:69351] 65 30 42 493 6 89 581 51 104 32 ...
 $ last_review                   : Date[1:69351], format: "2016-06-10" "2022-07-15" "2022-01-04" "2022-09-02" ...
 $ reviews_per_month             : num [1:69351] 0.47 0.2 0.34 3.59 0.05 0.57 4.26 0.36 0.86 0.29 ...
 $ calculated_host_listings_count: num [1:69351] 1 2 1 5 1 1 1 5 1 1 ...
 $ availability_365              : num [1:69351] 74 343 222 236 180 70 193 249 318 251 ...
...

Getting started with plotly

Import the plotly library
```
library(plotly)
```
A plotly object is defined by placing a ggplot object within ggplotly()
```
ggplotly(<ggplot object>)
```

Bubble plot with ggplot

Let's make a bubble plot and save it as bubble
- Show relationship between average price and availability_365
- Differently colored bubble for each room_type
- Size of bubble is proportional to the number of listings
- Add in additional layers for beautification

bubble <- listings %>%

  group_by(room_type, neighbourhood) %>%
  summarise(avg_price = mean(price, na.rm=TRUE),
            avg_availability = mean(availability_365, 
                                        na.rm=TRUE),

            Count = n(), .groups = "drop") %>%

  ggplot(aes(x = avg_price, y = avg_availability,

                  color = room_type,

                size = Count)) +

  geom_point() +

labs(x = "Average price", 
      y = "Average availability", 
    color = "Room type") +
  theme_classic()

Bubble plot with ggplot, rendered

A regular bubble plot that was created with ggplot.

Interactive bubble plot with plotly

Encase bubble in ggplotly()

ggplotly(bubble)

An interactive plotly bubble plot.

Horizontal boxplots with plotly

Some plots cannot be converted correctly
Create a series of horizontally oriented boxplots
- Describe availability_365 distribution with respect to room_type
Conversion with ggplotly() does not work

boxH <- listings %>%

  ggplot(aes(x = availability_365, y=room_type)) +

  geom_boxplot() +

  labs(x="Availbility (out of 365 days)", 
          y="Room type") + theme_classic()

ggplotly(boxH)

Horizontally oriented boxplots created with ggplot.

Horizontally oriented boxplots converted using ggplotly.

Vertical boxplots with plotly

Switch x and y in aes()
Plotly conversion will now work

boxV <- listings %>%

  ggplot(aes(y = availability_365, x=room_type)) +
  geom_boxplot() +
  labs(y="Availbility (out of 365 days)", x="Room type") + 
  theme_classic()

ggplotly(boxV)

Vertically oriented boxplots with ggplot.

Vertically oriented boxplots with ggplotly.

Fixing horizontal boxplots with plotly

What if we want horizonal box plots?
- Add coord_flip()
We can also use facet_wrap()
- Each panel must be large enough

ggplotly(boxV + coord_flip())

ggplotly(boxV + coord_flip() + 
    facet_wrap(~neighbourhood))

Flipped vertically oriented plotly boxplots.

Faceted plotly boxplots.

Let's practice!

Building Dashboards with shinydashboard