Dumbbell Plot

Homes Data

I spent two months house hunting before purchasing a home. The Homes Viewed.csv data set has 5 variables for each of the 29 homes:

Location: “Street Name, Town”
Type: House/Condo
Square Feet: The area of finished space (in square feet)
Asking Price: How much the sellers were asking when the house was listed.
Purchase Price: How much the buyers paid for the house.

Load the tidyverse package, read in the data, and skim() it to look at the data:

Data summary
Name	homes
Number of rows	29
Number of columns	5
_______________________
Column type frequency:
character	4
numeric	1
________________________
Group variables	None

Variable type: character

skim_variable	n_missing	complete_rate	min	max	n_unique
Address	0	1.00	13	25	29
Type	0	1.00	5	5	2
Asking.Price	0	1.00	4	4	24
Purchase.Price	1	0.97	4	4	21

Variable type: numeric

skim_variable	n_missing	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
Square.Feet	0	1	1751.31	338.43	1270	1533	1623	1935	2530	▅▇▅▃▂

A couple of issues we should fix:

The column names are not user friendly (Capitalized, spaces between words)
Purchase Price has a missing value
The 2 price variables are characters since the price have a “k” after them

Cleaing the data

So how do we clean the data to make it useable?

Use clean_names() from the janitor package to make the column names more user friendly

Installing the tidyverse package also installs janitor, but it doesn’t load it!

Use one of our main dplyr verbs to remove the house that has a missing Purchase Price
Use another one of our main dplyr verbs along with parse_number() to convert Asking Price and Purchase Price from a string to a numeric column.

“450k” –> 450000

We may want to have the name of the street and the name of the town in different columns. The separate() function in the tidyr package will separate a string into two (or more) different columns based on a separator character.

Use “,” to separate street and town in the Address column

separate() has three main arguments:

col = the column with the string to be divided - “Var1”
into = a pair of names to name the new columns - c("Var2", "Var3")
sep = a character used to divide the data into columns

Finally, name the new data set houses

##                street       town  type square_feet asking purchase
## 1           Venus Ave        NNE house        1407 360000   350000
## 2        Pearl Street      Essex house        1270 350000   360000
## 3           James Ave        NNE house        1528 325000   360000
## 4            Maple St       Bton house        1533 425000   400000
## 5            River St   Winooski condo        2263 380000   412000
## 6       Logwod Circle      Essex house        1620 370000   420000
## 7       Iroquis Drive      Essex house        2158 390000   425000
## 8           North Ave        NNE condo        1852 405000   427000
## 9           Arthur Ct       Bton condo        1579 390000   430000
## 10      Victory Drive         SB house        1796 396000   450000
## 11  Laurel Hill Drive         SB house        1935 432000   465000
## 12        Baldwin Ave         SB house        1555 430000   465000
## 13     Whitewater cir  Williston condo        2241 430000   475000
## 14         Thorton St   Winooski house        1554 440000   480000
## 15         Rudgate Rd Colchester house        2312 450000   511000
## 16           North St   Winooski house        1759 485000   527000
## 17 University Terrace       Bton house        1272 489000   550000
## 18      Devon Hill Ct      Essex house        2530 485000   550000
## 19   Juniper Ridge Rd      Essex house        1863 495000   552000
## 20        Colonial Dr Colchester house        1656 475000   590000
## 21          Foster St       Bton house        1489 519000   620000
## 22         Meridan St        NNE house        1824 410000   430000
## 23         Sundown Dr  Williston house        2270 435000   485000
## 24         Cascade St      Essex house        1623 418000   450000
## 25          Sherry Rd         SB house        1613 395000   441000
## 26        Lapointe St   Winooski house        1956 445000   480000
## 27      Lafountain St   Winooski house        1600 360000   360000
## 28       Saratoga Ave        NNE house        1428 425000   448000

Dumbbell Plot: Method 1

There is a direct way we can use to create a dumbbell plot with 1 major downside we’ll see when we finish making it.

Looking at the example dumbbell plot in Brightspace, for each house we need:

1 point for asking price
1 point for purchase price
A line to connect the two points

For the two points, we can use two separate geom_point() - one for each column.

What about the line that connects them? How do we draw a line that has a stopping and starting point?

With geom_segment(), which has 4 needed aesthetics that indicate the two sets of coordinates to start and stop the line segment:

starting coordinates: (x, y)
ending coordinates: (xend, yend)

Since we are drawing a flat line, the y and yend coordinates should be the same while x and xend will be the asking and purchase price. Try creating the basic graph:

What is the major problem?

That there isn’t a legend! The reader doesn’t know which color dot is the asking price vs the purchase price. So what can we do?

In order for ggplot() to automatically create a legend, the aesthetic that needs to be a guide needs to be mapped to a single column. So we need a column that has the price type (asking/purchase) and another that just has the prices, regardless if it is the asking or purchase price. So how do we do that?

Pivot verbs

If we want to create a dumbbell plot using ggplot, we need each of the variables in their own column so we can map them to their specific aesthetics:

y = case - house denoted by street
x = quantitative variable - house price
color = binary variable - asking or purchase price

Unfortunately, our data isn’t in the right format. The quantitative variable is in 2 different columns and the binary variable (price type) is represented by the column names.

If we want to use ggplot() to create a dumbbell plot, we need:

Both asking and purchase to be in the same column - price
An additional column that gives the type of price (asking vs purchase) in another column - price_type

This is called the long format of the data.

We can use pivot_longer() to wrangle the data into a long format for price. The arguments are:

cols = The columns we want to “stack” on top of one another
names_to = the column name for the additional column (that contains the old column names)
values_to = the column name for the column that stores the stacked columns

Both the names_to and values_to arguments need the new names to be in quotes - “var1”

Use pivot_longer() to manipulate the data to be in the long format and call it houses_long

## # A tibble: 56 × 6
##    street       town     type  square_feet price_type  price
##    <chr>        <chr>    <chr>       <int> <chr>       <dbl>
##  1 Venus Ave    NNE      house        1407 asking     360000
##  2 Venus Ave    NNE      house        1407 purchase   350000
##  3 Pearl Street Essex    house        1270 asking     350000
##  4 Pearl Street Essex    house        1270 purchase   360000
##  5 James Ave    NNE      house        1528 asking     325000
##  6 James Ave    NNE      house        1528 purchase   360000
##  7 Maple St     Bton     house        1533 asking     425000
##  8 Maple St     Bton     house        1533 purchase   400000
##  9 River St     Winooski condo        2263 asking     380000
## 10 River St     Winooski condo        2263 purchase   412000
## # ℹ 46 more rows

Now we have the data in the format we need!

Dumbbell plot

Now use ggplot(), geom_point(), and geom_line() to create a dumbbell plot!

To match the graph in blackboard, set size to 3 in geom_point() and 1 in geom_line()

Let’s go through a couple of ways we can improve the graph as a class!

Dumbbell Plot

Jacob Martin

DS 187 - Chapter 5

What is a Dumbbell Plot?

Homes Data

Cleaing the data

Dumbbell Plot: Method 1

Pivot verbs

Dumbbell plot