A course in quantitative research workflow for students in the higher education administration program at the University of Florida
In this supplemental lesson, we’ll make a few interactive plots using the plotly library. Why might you want to make interactive plots? Aside from the fact that they are very cool, they can help you make a stronger data presentation.
How so?
First, the wow! factor of interactive plots can bring people in. Rather than passively reading a report or watching you present your findings, your audience can actively participate in their own understanding. Even if your interactive plot isn’t any special, the process of a person using/”playing” with it means you’ve got their attention, at least for a moment. A well designed plot, like a well-designed interactive exhibit at a museum, can use that attention to instruct.
Second, interactive features allow you to plot along new dimensions. With a static 2D graph, we start hitting a wall at about 4 dimensions at once: x-axis, y-axis, color/shape, small multiples, etc. If we are smart about it, we can fit a lot of information into one figure, but it’s difficult. With interactive features, we can add a 3rd dimension, dynamic visuals that change over time, tooltips to show point-specific information, and buttons that allow the user to change the underlying data. Interactive graphics can allow for more options since they can be changed on the fly.
The good news is that plotly in R works very similarly to ggplot. In fact, it’s often trivial to convert a static ggplot figure into an interactive plotly figure. In this lesson, however, I’ll show you how to use plotly’s particular interface, which is a little different.
As a note, most of what I’m showing you here is just code from the R plotly website that I’ve modified to replicate some figures we’ve made before. As usual, what I’m showing you only scratches the surface of what’s possible. If you want to make different types of interactive figures or modify what you see here, check out the plotly web page: plotly.com/r/.
NOTE I’ve included a PDF link at the top of this page as always, but interactive plots don’t play well with PDFs so you won’t see the figures there. Be sure to run the code in RStudio or use the website to see the interactive figures.
Unless you have already done so, don’t forget to install plotly
library using install.packages("plotly").
## ---------------------------
## libraries
## ---------------------------
library(tidyverse)
## ── Attaching packages ─────────────────────────────────────── tidyverse 1.3.1 ──
## ✔ ggplot2 3.3.5 ✔ purrr 0.3.4
## ✔ tibble 3.1.6 ✔ dplyr 1.0.8
## ✔ tidyr 1.2.0 ✔ stringr 1.4.0
## ✔ readr 2.1.2 ✔ forcats 0.5.1
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
library(haven)
library(plotly)
##
## Attaching package: 'plotly'
## The following object is masked from 'package:ggplot2':
##
## last_plot
## The following object is masked from 'package:stats':
##
## filter
## The following object is masked from 'package:graphics':
##
## layout
As we did with the the lesson on making
graphics, we’ll use two sets of data:
hsls_small.dta and all_schools.csv.
Note that since we have two data files this lesson, we’ll give
them unique names instead of the normal df:
df_hs := hsls_small.dtadf_ts := all_schools.csvAnd as always, we are working in the ./scripts subdirectory.
## ---------------------------
## directory paths
## ---------------------------
## assume we're running this script from the ./scripts subdirectory
dat_dir <- file.path("..", "data")
tsc_dir <- file.path(dat_dir, "sch_test")
## ---------------------------
## input data
## ---------------------------
## assume we're running this script from the ./scripts subdirectory
## read_dta() ==> read in Stata (*.dta) files
## read_csv() ==> read in comma separated value (*.csv) files
df_hs <- read_dta(file.path(dat_dir, "hsls_small.dta"))
df_ts <- read_csv(file.path(tsc_dir, "all_schools.csv"))
## Rows: 24 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): school
## dbl (4): year, math, read, science
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
Pretty much whatever plot you can make in ggplot(), you can make using plotly. We’ll start with histograms to get our bearings.
Just like ggplot starts with the function ggplot(), plotly starts
with plot_ly() (notice the underscore in the function name). Like
you have seen with ggplot(), plot_ly() starts with the data you want
to use.
Unlike ggplot(), you don’t use an aesthetics—aes()—function to
map the data to axes. Instead, you set the values directly, but use a
tilde (~) next to the variable name. Think of the tilde as saying,
I want this variable to be interactive in some way. For our first
histogram, we’re going to plot the distribution of math test scores:
~x1txmtscor.
Also unlike ggplot(), we set the figure type (geom in ggplot-speak)
inside the plot_ly() function using the argument type. In this
case, type = "histogram").
We could just call plot_ly() and let the plot pull up in our plot
window, but we’ll first save it in an object, p. We’ll then call
p.
One thing to notice about plotly figures is that they are *.html
files by default. This is the markup language for websites
(Hyper Text Markup Language). If you are using
RStudio, the plot should pull up in your plot viewer. Outside of
RStudio, it will open you in your web browser.
## create basic histogram with plotly
p <- plot_ly(data = df_hs, x = ~x1txmtscor, type = "histogram")
## show
p
This first histogram is pretty plain, but run your mouse over it. You
should notice a tooltip next to your mouse. There are
two numbers. These represent the x-axis range of the histogram bin and
the number of observations falling into that bin (y-axis). So if you
see, (49.5 - 50, 571), that’s telling you that 571 observations have
test scores within 49.5 and 50. With ggplot(), we don’t get that level
of detail.
You should also see a number of controls on the upper part of the figure. These allow you to zoom in and and save the figure. If you click and drag on the figure, you can zoom in on a certain region. Take a few moments to play around with the figure and controls.
If we want to show a density histogram rather than a frequency
histogram, we only need to add the option histnorm = "probability"
as an argument. All else can stay the same.
## create histogram plotly
p <- plot_ly(data = df_hs,
x = ~x1txmtscor,
histnorm = "probability",
type = "histogram")
## show
p
Notice that the y-axis is now on the probability scale. Similarly, the tooltip now shows the probability of the bin rather than the count of observations that fall into it.
QUICK EXERCISE
Create an interactive histogram using another continuous variable in the
hsls_smalldata set. You can make either a frequency or density histogram.
Now that we’ve made a couple, let’s make a nicer looking
histogram. First, within the plot_ly() function, we’ll add the
argument marker and some options to make our histogram bars look a
little different. One thing to notice about plotly options is that
they tend to fall into lists. This is likely due to the fact that
plotly works with other programming languages like Python and is
splitting the difference between idioms. With marker we add two
named list items: color and line, which also takes a list. With
these options, we can change the fill (inside) color of the bars and
edge color and line thickness.
Next, we add nicer labels using the layout() function. Notice that
unlike ggplot, which connects items with a plus (+) sign, plotly
uses the magrittr pipe (%>%). Inside layout(), we set the overall
title and the x-axis label (which also use the title name).
Finally, we improve the tooltip. With argument hovertemplate we
paste() together a new string. Anything static in the string, like
"Bin width:" will stay static in the tooltip. So the values for bin
width and probability change, we include the variables inside
%{<var>}. For example, notice that next to "Bin width: " we
include %{x}. This means that that value should change along values
of x, our test score. It’s a little tricky here since we don’t
actually have a y value that we set explicitly. That said, we know
that y is probability in this histogram. So that we don’t have a
redundant name on our tooltip, we set name = "".
## create histogram plotly w/
## 1. better labels
## 2. add title
## 3. change color of bars
## 4. add outline to bars
## 5. improve the tooltip
p <- plot_ly(data = df_hs,
x = ~x1txmtscor,
histnorm = "probability",
type = "histogram",
marker = list(color = "#E28F41",
line = list(color = "#6C9AC3",
width = 2)),
name = "",
hovertemplate = paste("Bin width: %{x}",
"<br>", # add HTML <br> for line break
"Probability: %{y}")) %>%
layout(title = "Distribution of math test scores",
xaxis = list(title = "Math test score"))
## show
p
That’s much nicer looking! Notice how much nicer and more user friendly the tooltip is now, too.
Plotting the difference in a continuous distribution across groups is
a common task. As we’ve done before, let’s see the difference between
student math scores between students with parents who have any
postsecondary degree and those without. Since we’re using data that
was labeled in Stata, we’ll see the labels when we use count().
## see the counts for each group
df_hs %>% count(x1paredu)
## # A tibble: 7 × 2
## x1paredu n
## <dbl+lbl> <int>
## 1 1 [Less than high school] 1010
## 2 2 [High school diploma or GED] 5909
## 3 3 [Associate's degree] 2549
## 4 4 [Bachelor's degree] 4102
## 5 5 [Master's degree] 2116
## 6 7 [Ph.D/M.D/Law/other high lvl prof degree] 1096
## 7 NA 6721
Now that we see our options, we’ll make a new variable, pared_coll
that equals 1 if either of the student’s parents have any college
degree and 0 otherwise. We’ll store this smaller data frame in a new
object, plot_df, that we’ll us to plot a new two-way histogram.
## need to set up data
plot_df <- df_hs %>%
## select the columns we need
select(x1paredu, x1txmtscor) %>%
## can't plot NA so will drop
drop_na() %>%
## create new variable that == 1 if parents have any college
mutate(pared_coll = ifelse(x1paredu >= 3, 1, 0)) %>%
## drop (using negative sign) the original variable we don't need now
select(-x1paredu)
## show
plot_df
## # A tibble: 16,429 × 2
## x1txmtscor pared_coll
## <dbl+lbl> <dbl>
## 1 59.4 1
## 2 47.7 1
## 3 64.2 1
## 4 49.3 1
## 5 62.6 1
## 6 58.1 1
## 7 49.5 0
## 8 54.6 1
## 9 53.2 0
## 10 63.8 1
## # … with 16,419 more rows
Unlike with ggplot, we won’t just add a group = ... aesthetic to the
plot. Instead, we’ll add two separate histograms, one for each
pared_coll group and then tell plotly to barmode = "overlay" in
the layout() function. To only plot one pared_coll group at a
time, notice that we use plot_df %>% filter(pared_coll == ...) in
each data argument, replacing ... with 0 and 1.
We include hovertemplate arguments in each add_histogram()
function. We also include redundant text and name arguments. With
the text argument, we can include the pared_coll status in the
tooltip with %{text}. With name, the legend will be properly
labeled. However, adding name back in means that we get that
annoying extra label on our tooltip. To remove it, we add
"<extra></extra>" at the end of our hovertemplate.
## two way histogram: add one at a time
p <- plot_ly(alpha = 0.5) %>%
## first: when parents don't have college degree
add_histogram(data = plot_df %>% filter(pared_coll == 0),
x = ~x1txmtscor,
histnorm = "probability",
type = "histogram",
name = "No college",
hovertemplate = paste("No college",
"<br>",
"Bin width: %{x}",
"<br>",
"Probability: %{y}",
"<extra></extra>")) %>%
## second: when parents have college degree
add_histogram(data = plot_df %>% filter(pared_coll == 1),
x = ~x1txmtscor,
histnorm = "probability",
type = "histogram",
name = "Some college or more",
hovertemplate = paste("Some college or more",
"<br>",
"Bin width: %{x}",
"<br>",
"Probability: %{y}",
"<extra></extra>")) %>%
## tell plotly to overlay the histograms
layout(barmode = "overlay",
title = "Distribution of math test scores",
xaxis = list(title = "Math test score"),
legend = list(title = list(text = "Parent's education level")))
## show
p
QUICK EXERCISE
Create an interactive two-way histogram using another grouping variable in the
hsls_smalldata set (something other than parental college). You can use an existing indicator variable or create your own.
Here we’ll replicate the box plot that we’ve made before. Since the
xaxis boxes and the different colors represent the same things—a
different level of parental educational expectation—we’ll turn off
the legend with showlegend = FALSE.
## box plot using as_factor()
p <- plot_ly(data = df_hs,
color = ~as_factor(x1paredu),
y = ~x1txmtscor,
type = "box") %>%
layout(title = "Math score by parental expectations",
xaxis = list(title = "Parental expectations"),
yaxis = list(title = "Math score"),
showlegend = FALSE)
## show
p
Nothing too fancy here. Our factor labels are a bit long, so we could
probably make a cleaner figure if shortened them. The biggest benefit
of the interactive boxplot is that the tooltip shows the values of the
various levels: min, lower fence, Q1, median, Q3, upper
fence, and max. That said, it’s a bit busy, so you’ll have to
decide whether the benefit outweighs the cost.
Interactive scatter plots can be really useful if for no other reason than they allow you to zoom in and pick out specific values, which is great for identifying outliers.
As with a prior lesson, we’ll limit our data to only a 10% sample. Remember that because it’s a random process, your data and resulting figures will look a little different from mine.
## sample 10% to make figure clearer
df_hs_10 <- df_hs %>%
## drop observations with missing values for x1stuedexpct
drop_na(x1stuedexpct) %>%
## sample
sample_frac(0.1)
To make the scatter plot, we’ll start by setting type = "scatter"
and mode = "markers".
## scatter
p <- plot_ly(data = df_hs_10,
x = ~x1ses,
y = ~x1txmtscor,
type = "scatter",
mode = "markers")
## show
p
Like the other plotly figures, we can hover over specific points to
get their values. Let’s do another version, but cleaned up: better
axes, better tooltip, and better colors. One new thing: we set
zeroline = FALSE in the layout() function. This prevents the
figure from drawing a bold vertical line at 0. You can remove the
option and have the bold line if you want, but it seemed unnecessary
here so I’ve dropped it.
## scatter
p <- plot_ly(data = df_hs_10,
x = ~x1ses,
y = ~x1txmtscor,
type = "scatter",
mode = "markers",
## set point options
## size: radius
## color: fill color
## line: circumference line width and color
marker = list(size = 10,
color = "#E28F41",
line = list(color = "#6C9AC3",
width = 2)),
## turn off side name on hover
name = "",
hovertemplate = paste("SES: %{x}",
"<br>", # add <br> for line break
"Math: %{y}")) %>%
layout(title = "Math scores as function of SES",
xaxis = list(title = "SES",
zeroline = FALSE),
yaxis = list(title = "Math score"))
## show
p
Let’s add another dimension: color representing whether the student
planned to graduate from college or not. First, we’ll check the levels
of our key variable, x1stuedexpct).
## see student base year plans
df_hs %>%
count(x1stuedexpct)
## # A tibble: 12 × 2
## x1stuedexpct n
## <dbl+lbl> <int>
## 1 1 [Less than high school] 93
## 2 2 [High school diploma or GED] 2619
## 3 3 [Start an Associate's degree] 140
## 4 4 [Complete an Associate's degree] 1195
## 5 5 [Start a Bachelor's degree] 115
## 6 6 [Complete a Bachelor's degree] 3505
## 7 7 [Start a Master's degree] 231
## 8 8 [Complete a Master's degree] 4278
## 9 9 [Start Ph.D/M.D/Law/other prof degree] 176
## 10 10 [Complete Ph.D/M.D/Law/other prof degree] 4461
## 11 11 [Don't know] 4631
## 12 NA 2059
We see that x1stuedexpct >= 6 means a student plans to earn a
Bachelor’s degree or higher. But since we need to account for the fact
that 11 means “I don’t know”, we need to make sure our test includes
x1stuedexpct < 11. Remember from a prior lesson that we can connect
these two statements together with the operator &. Let’s create our
new variable.
## create variable for students who plan to graduate from college
df_hs_10 <- df_hs_10 %>%
mutate(plan_col_grad = ifelse(x1stuedexpct >= 6 & x1stuedexpct < 11,
"Yes", # if T: "Yes"
"No")) # if F: "No"
Now that we have our new variable plan_col_grad, we can add it the
color aesthetic. How do we specify the colors? Notice that we
created an object called pal with two Hex color codes. In the next
line, we named these colors in the vector with “Yes” and “No” to match
our new plan_col_grad variable. We add this to the colors
argument. That will assign the first color in pal to all the Yess and
the second color in pal to all the Nos.
This time, we’re also making our scatterplot a little
differently. We’re using add_trace() this time. Most of the
arguments are the same as before, but are in the add_trace()
function rather than in the plot_ly() function.
## set color palette with names
pal <- c("#E28F41","#6C9AC3")
pal <- setNames(pal, c("Yes", "No"))
## scatter plot
p <- plot_ly() %>%
add_trace(data = df_hs_10,
x = ~x1ses,
y = ~x1txmtscor,
color = ~plan_col_grad, # color changes by college plans
colors = pal, # using pal from above
type = "scatter",
mode = "markers",
hovertemplate = paste("SES: %{x}",
"<br>",
"Math: %{y}",
"<extra></extra>")) %>%
layout(title = "Math scores as function of SES",
xaxis = list(title = "SES",
zeroline = FALSE), # turn off bold zero line
yaxis = list(title = "Math score"),
legend = list(title = list(text = "Plans to graduate from college?")))
## show
p
There’s quite a bit of overlap, but we can see that students who plan to graduate from college tend to have a higher SES and math scores.
Let’s take full advantage of plotly and make a 3D plot. It would be ideal if we had a third continuous variable to plot against, but with our limited data set, we’ll settle for the number of months between HS graduation and first college enrollment on the z axis.
To add the third dimension repeats most of the same code. We’re back
using plot_ly() for most of the set up, adding z = ~x4hs2psmos and
a new line to our tooltip. We call the points with the add_markers()
function.
Labeling the axes is slightly different. For a 3D plot, all the x, y,
and z-axis labels need to go inside the scene argument list. But
other than that, the labels are the same.
## set color palette with names
pal <- c("#E28F41","#6C9AC3")
pal <- setNames(pal, c("Yes", "No"))
## scatter along three dimensions; most settings inside plot_ly() function now
p <- plot_ly(data = df_hs_10,
x = ~x1ses,
y = ~x1txmtscor,
z = ~x4hs2psmos,
color = ~plan_col_grad,
colors = pal,
## add extra row to tool tip
hovertemplate = paste("SES: %{x}",
"<br>",
"Math: %{y}",
"<br>",
"HS to College: %{z} months",
"<extra></extra>"),
## make marker a little smaller
marker = list(size = 3)) %>%
## now tell plot_ly to add points as markers
add_markers() %>%
## in 3D, set axis titles inside scene() argument
layout(title = "Math scores as function of SES",
scene = list(xaxis = list(title = "SES"),
yaxis = list(title = "Math score"),
zaxis = list(title = "Months between HS and college")),
legend = list(title = list(text = "Plans to graduate from college?")))
## show
p
Click on the figure and move it around. You’ll see that in addition to zooming in and out, you can rotate the figure to better see the relationship across the three variables. For this particular example, the relationship is difficult to see (3D figures aren’t always the best option!). Most students who enrolled in college did so pretty soon after graduating high school. Just by looking, there doesn’t seem to be much difference between the two groups (though some more formal testing might be in order).
QUICK EXERCISE
Create an interactive 3D scatter plot using SES, math scores, and a categorical variable with more than two groups. This means your legend should more than two items and your graph should have more than two colors. Note that you likely need to increase the number of colors in
pal(you can find more hex codes at https://htmlcolorcodes.com) or remove the argument altogether.
We can also make line graphs with plotly. For these, we’ll once again use our school test score data. We won’t go through all the iterations in the first graphing lesson, but know that you can convert those to interactive figures as well.
We’ll begin by showing our data (which is wide-ish) and creating a fully long version of the data.
## show test score data
df_ts
## # A tibble: 24 × 5
## school year math read science
## <chr> <dbl> <dbl> <dbl> <dbl>
## 1 Bend Gate 1980 515 281 808
## 2 Bend Gate 1981 503 312 814
## 3 Bend Gate 1982 514 316 816
## 4 Bend Gate 1983 491 276 793
## 5 Bend Gate 1984 502 310 788
## 6 Bend Gate 1985 488 280 789
## 7 East Heights 1980 501 318 782
## 8 East Heights 1981 487 323 813
## 9 East Heights 1982 496 294 818
## 10 East Heights 1983 497 306 795
## # … with 14 more rows
## reshape data long (as we've done in a prior lesson)
df_ts_long <- df_ts %>%
pivot_longer(cols = c("math","read","science"), # cols to pivot long
names_to = "test", # where col names go
values_to = "score") %>% # where col values go
group_by(test) %>%
mutate(score_std = (score - mean(score)) / sd(score)) %>%
group_by(test, school) %>%
arrange(year) %>%
mutate(score_year_one = first(score),
## note that we're using score_year_one instead of mean(score)
score_std_sch = (score - score_year_one) / sd(score)) %>%
ungroup
## show
df_ts_long
## # A tibble: 72 × 7
## school year test score score_std score_year_one score_std_sch
## <chr> <dbl> <chr> <dbl> <dbl> <dbl> <dbl>
## 1 Bend Gate 1980 math 515 1.40 515 0
## 2 Bend Gate 1980 read 281 -0.863 281 0
## 3 Bend Gate 1980 science 808 0.759 808 0
## 4 East Heights 1980 math 501 0.115 501 0
## 5 East Heights 1980 read 318 1.34 318 0
## 6 East Heights 1980 science 782 -0.735 782 0
## 7 Niagara 1980 math 514 1.31 514 0
## 8 Niagara 1980 read 292 -0.208 292 0
## 9 Niagara 1980 science 787 -0.448 787 0
## 10 Spottsville 1980 math 498 -0.161 498 0
## # … with 62 more rows
First, we’ll only plot one school, Bend Gate. As we did above, we
won’t add a grouping argument for each test. Instead, we’ll add a
unique add_trace() for each test—math, reading, and
science—which are added to the y argument. In the first
add_trace(), we’ll include our wide data and filter(school == "Bend
Gate"). We don’t need to include data in the other functions since it
will be included by default. We set type = "scatter" and mode =
"lines" to make our line graphs with x = ~year. Finally, we add the
argument hovermode = "x unified" to the layout() function, which
connects our line information in the tooltip.
## scatter plot
p <- plot_ly() %>%
## add data for Bend Gate only; y == math
add_trace(data = df_ts %>% filter(school == "Bend Gate"),
x = ~year,
y = ~math,
name = "Math",
type = "scatter",
mode = "lines") %>%
## repeated, but this time y == reading
add_trace(x = ~year,
y = ~read,
name = "Reading",
type = "scatter",
mode = "lines") %>%
## repeated, but this time y == science
add_trace(x = ~year,
y = ~science,
name = "Science",
type = "scatter",
mode = "lines") %>%
## add unified hovermode so that tooltip for all test scores pops ups
layout(title = "Test scores at Bend Gate: 1980 - 1985",
xaxis = list(title = "Year"),
yaxis = list(title = "Score"),
legend = list(title = list(text = "Test")),
hovermode = "x unified")
## show
p
Notice how the tooltip includes information for all tests in the same
year, no matter which line you hover over? That’s due to the
hovermode = "x unified" argument. You can omit that argument and the
tooltip will work as normal (local to the point under your mouse
pointer), but in this case, it makes sense to link them.
How to add all the school values in one plot like we did with ggplot’s
facet_wrap()? It’s a little trickier, but we can do it with a loop
(see the lesson on functional
programming for a review of loops).
First, we’ll store our school names in an object, schools. Next,
we’ll initialize a blank list with plot_list <- list() that will
store our school-specific plots. Inside the loop, we’ll filter the
data with filter(school == i) where i will equal whichever school
we’re on in the loop: Bend Gate the first iteration, East Heights the
second, and so on.
So that we don’t have repeated legends in the final figure, we do a
few things. First, we add the argument legendgroup = ~test so that
plotly knows the tests (math, reading, and science) are grouped
together within school. Next, we set showlegend = if_first. What’s
if_first? It’s a Boolean (TRUE/FALSE) that’s only TRUE on the
first loop iteration, when i == 1, and FALSE otherwise. This means
that we only include a legend with the first school figure. Since
we’re going to put them all together at the end, that one will suffice
for all of them. Finally, we use add_annotations() with various
options so that the name of each school is printed on the top if its
own figure (text = i).
Once the loop has run, we use the function subplot() to join the
plots into one figure. With argument nrows = 2, we’ll end up with a
2x2 grid.
## set up vector of school names
schools <- c("Bend Gate", "East Heights", "Niagara", "Spottsville")
## init list to hold plots
plot_list <- list()
## loop through schools to make each plot at a time
for (i in schools) {
## TRUE if first school, otherwise FALSE;
## use so we only end up with one legend
if_first <- (i == 1)
## store scatter plot in list using index i
plot_list[[i]] <- plot_ly() %>%
## filter to school i (one school at a time)
add_trace(data = df_ts_long %>% filter(school == i),
x = ~year,
y = ~score_std_sch,
color = ~test,
legendgroup = ~test,
text = ~score,
showlegend = if_first, # only TRUE the first time
type = "scatter",
mode = "lines",
## notice that we include scaled and actual score in hover
hovertemplate = paste("Year: %{x}",
"<br>",
"Score (scaled): %{y}",
"<br>",
"Score (actual): %{text}")) %>%
layout(xaxis = list(title = "Year"),
yaxis = list(title = "Score"),
legend = list(title = list(text = "Test"))) %>%
## settings to add school name to title of each subplot
add_annotations(text = i,
x = 0,
y = 1,
yref = "paper",
xref = "paper",
xanchor = "middle",
yanchor = "top",
showarrow = FALSE,
font = list(size = 15))
}
## combine subplots into on main plot like ggplot() facet_wrap()
p <- subplot(plot_list[[1]], plot_list[[2]], plot_list[[3]], plot_list[[4]],
nrows = 2)
## show
p
Okay! The figure is a little scrunched on my website, but it might look better by itself on a larger monitor. We could also play with some sizing, but since HTML is sized on the fly and unique to each screen, that may not be worth our time.
What is particularly cool about this figure is that even though we plot the first-year standardized scores, we can include the real scores in the tooltip. That let’s us have the best of both worlds: the line graph shows the relationship we care about (how have scores changed relative to each other and the first year?), but have the real scores at our finger tips.
Lastly, we can also use plotly to make a table. While perhaps not as exiting as the figures, the table is interactive in that it prints pretty (includes a scroll bar) and that the user can drag to reorder columns. As a demo, we’ll make a table using the raw test score data underlying the line graph we just made.
To make a table, you need to three key components:
df_ts)For 1 and 3, most of the formatting is simply to change the background
and font colors. For the values in the header list, we’ll use the
names from our wide-ish data frame. So that they are capitalized,
we’ll use str_to_title(), which capitalizes the first letter
only: str_to_title(names(df_ts)).
For the data in the cells, well convert our data frame to a matrix
using as.matrix(). Because of a quirk of the program, we’ll need to
transpose the matrix with t(). Finally, so we don’t repeat the
column names as the first (remember, we’ve already included them in
header), we’ll unname() the data frame. All together, it’s
t(as.matrix(unname(df_ts))).
## make interactive table
tab <- plot_ly(
type = "table",
## adjust how header row looks
header = list(
## use str_to_title() with tibble column names: math --> Math
values = c(str_to_title(names(df_ts))),
## left align school names, and center all other columns (hence -1)
align = c("left", rep("center", ncol(df_ts) - 1)),
## make vertical lines thicker
line = list(width = 1, color = "black"),
## fill color with blue
fill = list(color = "rgba(108, 154, 195, 0.8)"),
## set font to sans serif with bigger size and white color
font = list(family = "Arial", size = 14, color = "white")),
## adjust how cells look
cells = list(
## use df_ts, but...
## 1. convert to matrix,
## 2. transpose using t(),
## 3. drop names (already in header row)
values = t(as.matrix(unname(df_ts))),
## same alignment as header
align = c("left", rep("center", ncol(df_ts) - 1)),
## same vertical line setup as header
line = list(color = "black", width = 1),
## fill first column different color to stand out
fill = list(color = c("rgba(108, 154, 195, 0.5)",
"rgba(226, 143, 65, 0.5)")),
## same font as header, but smaller and different color
font = list(family = "Arial", size = 12, color = c("black"))
))
## show
tab
If you hover over a cell and then click, you should be able to drag and drop the columns into a new order. It seems a bit much for this small data set, but having an “active” data table that you can manipulate could come in handy during meetings or your own research process.