R-basic-dataviz.md

title	output
Basic data visualization in R	html_document keep_md true

By Christine Zhang (ychristinezhang at gmail dot com and @christinezhang on Twitter)
Data+Narrative Workshop at Boston University (June 7, 2018)

GitHub repository for Data+Code: https://github.com/underthecurve/r-dataviz-ggplot2

Data visualization is a popular aspect of R, since R can create sophisticated graphics with (relatively few) lines of code. In this class, we'll go through the basic data visualization functions of R.

In this workshop, we will:

• Go through a quick overview of some common graphic types
• Plot some common chart types in R

Overview of common graphic types

What are some common chart types?

• bar plots
• line plots
• scatterplots
• histograms
• boxplots

When do you think you should (or should not) use each of these chart types?

The Financial Times Visual Vocabulary is a good guide to different chart types and how to use them: https://github.com/ft-interactive/chart-doctor/tree/master/visual-vocabulary. The Data Visualisation Catalogue by Severino Ribecca is another one: https://datavizcatalogue.com/index.html. Basically all the resources listed under this blog post by Maarten Lambrechts are great: https://xeno.graphics/articles/on-graphonyms-the-importance-of-chart-type-names/.

In this class, we'll mainly use R's built-in graphics tool to take a look at the above common chart types. The next class, "Advancing with data visualization in R using ggplot2", will provide an overview of ggplot2 (which stands for the "grammar of graphics"), a popular R package for visualizing data.

Loading tidyverse

First, let's load the tidyverse package, which is a collection of R packages designed for doing data science.

# install.packages('tidyverse') # if you don't already have tidyverse
library('tidyverse') # load the tidyverse package

Today is June 7th, 2018. Let's take a look at the Boston weather on June 7th over the past decade or so.

Let's load in a data file, boston_weather.csv, of hourly Boston weather from 2007-2017, using the read_csv() command (part of the tidyverse). These historical weather figures have been obtained from Weather Underground. For more information on how I got the data from the website, see the processing-weather-data.R file. We'll call the dataframe boston.weather.

boston.weather <- read_csv('boston_weather.csv')

## Parsed with column specification:
## cols(
##   time = col_time(format = ""),
##   temp = col_double(),
##   dewpoint = col_double(),
##   windspeed = col_double(),
##   precip = col_double(),
##   events = col_character(),
##   conditions = col_character(),
##   year = col_integer(),
##   month = col_integer(),
##   day = col_integer()
## )

We can take a look at the first six rows of boston.weather using head():

head(boston.weather)

## # A tibble: 6 x 10
##   time    temp dewpoint windspeed precip events conditions     year month
##   <time> <dbl>    <dbl>     <dbl>  <dbl> <chr>  <chr>         <int> <int>
## 1 00:54   54       39.9       8.1     NA <NA>   Clear          2007     6
## 2 01:54   53.1     39.9       9.2     NA <NA>   Partly Cloudy  2007     6
## 3 02:54   53.1     41         9.2     NA <NA>   Clear          2007     6
## 4 03:54   52       41         8.1     NA <NA>   Clear          2007     6
## 5 04:54   51.1     41         9.2     NA <NA>   Clear          2007     6
## 6 05:54   52       41         9.2     NA <NA>   Clear          2007     6
## # ... with 1 more variable: day <int>

Bar plot

The boston.weather data provides hourly temperatures (note these are in 24-hour time and degrees Fahrenheit, respectively) for each year's June 7th. You can take a look for yourself on the Weather Underground webpage. For instance, here is the site URL for June 7, 2007:

https://www.wunderground.com/history/airport/KBOS/2007/6/7/DailyHistory.html?req_city=&req_state=&req_statename=&reqdb.zip=&reqdb.magic=&reqdb.wmo=

Here's a simple way to make a bar plot of the temp variable in the boston.weather -- every temperature in the data.

barplot(boston.weather$temp)

This is okay for a quick look, but it's not very informative. Each hour is on the x-axis and the temperature for that hour is on the y-axis. The data are plotted in the order that they appear in the dataframe.

Instead, let's plot just the highest temperature for each year. We will use the dplyr package (also included in the tidyverse) to do this. dplyr is a great tool for cleaning data in R. If you'd like to learn more about the syntax of dplyr, here are the materials from my workshop last year, "Tricks for cleaning your data in R": https://github.com/underthecurve/r-data-cleaning-tricks

Let's call this new dataframe with the maximum temperatures between 2007 and 2017 boston.weather.max.

boston.weather.max <- boston.weather %>% # the dataframe
  group_by(year) %>% # the grouping variable
  summarise(max.temp = max(temp)) # the variable we want: maximum temperature

We can take a look at the first six rows of boston.weather.max using head():

head(boston.weather.max)

## # A tibble: 6 x 2
##    year max.temp
##   <int>    <dbl>
## 1  2007     70  
## 2  2008     91.9
## 3  2009     82  
## 4  2010     73  
## 5  2011     82  
## 6  2012     63

Now let's try the barplot command again.

barplot(boston.weather.max$max.temp)

Notice how there are no x-axis labels, which is not particularly helpful in this case. We can add them using names.arg = :

barplot(boston.weather.max$max.temp, 
        names.arg = boston.weather.max$year)

If you're like me, you may be bothered by the fact the y-axis labels end at 80, but the temperatures in the data don't. For example, in 2008, the max temperature was 91.9. We can adjust the y-axis limits using ylim = . Let's make the y-axis go up to 100:

barplot(boston.weather.max$max.temp, 
        names.arg = boston.weather.max$year, 
        ylim = c(0, 100))

Line plot

Since this is time trend data, a line rather than a bar plot might be a better way of visualizing it. We can use the plot() function to do this:

?plot()
plot(x = boston.weather.max$year, 
     y = boston.weather.max$max.temp, 
     type = 'l')

Let's add axis labels and a title.

plot(x = boston.weather.max$year, 
     y = boston.weather.max$max.temp, 
     type = 'l', 
     xlab = 'year',
     ylab = 'temperature (in F)',
     main = 'Maximum temperatures in Boston on June 7th\n2007-2017\n')

Line plots are nice because we can visualize multiple things on one chart. For example, let's create another dataframe, boston.weather.min, again using a set of functions powered by dplyr. This dataframe will include the lowest, or minimum, temperatures for each year.

boston.weather.min <- boston.weather %>% # the dataframe
  group_by(year) %>% # the grouping variable
  summarise(min.temp = min(temp)) # the variable we want: minimum temperature

How can we inspect the first few rows of the boston.weather.min dataframe?

Now let's add a line plot of minimum temperatures to our plot of maximum temperatures. Highlight the following code and run it all at once:

plot(x = boston.weather.max$year, 
     y = boston.weather.max$max.temp,
     type = 'l', 
     xlab = 'year',
     ylab = 'temperature (in F)',
     main = 'Max and min temperatures in Boston on June 7th\n2007-2017\n')
lines(x = boston.weather.min$year, 
      y = boston.weather.min$min.temp)

This plot is ... not great. Why not?

Let's fix the y-axis limits first. How can we decide what these limits should be?

Highlight the following code and run it all at once:

plot(x = boston.weather.max$year, 
     y = boston.weather.max$max.temp,
     type = 'l', 
     xlab = 'year',
     ylab = 'temperature (in F)',
     main = 'Max and min temperatures in Boston on June 7th\n2007-2017\n',
     ylim = c(40, 100))
lines(x = boston.weather.min$year, 
      y = boston.weather.min$min.temp)

Maybe we also want the max and min to be different colors. We can do this with col. Here's a guide to how colors are named in R: http://www.stat.columbia.edu/~tzheng/files/Rcolor.pdf

Highlight the following code and run it all at once:

plot(x = boston.weather.max$year, 
     y = boston.weather.max$max.temp,
     type = 'l', 
     xlab = 'year',
     ylab = 'temperature (in F)',
     main = 'Max and min temperatures in Boston on June 7th\n2007-2017\n',
     ylim = c(40, 100), 
     col = 'red')
lines(x = boston.weather.min$year, 
      y = boston.weather.min$min.temp,
      col = 'blue')

We can also add a legend to our plot but this is a bit cumbersome in base R, so I'd encourage you to stick around for the next class, "Advancing with data viz in R using ggplot2": https://github.com/underthecurve/r-dataviz-ggplot2/blob/master/R-advancing-with-dataviz.md

For now, let's take a look at the legend() function:

?legend()

What should the arguments inside of legend() be?

Highlight the following code and run it all at once:

plot(x = boston.weather.max$year, 
     y = boston.weather.max$max.temp,
     type = 'l', 
     xlab = 'year',
     ylab = 'temperature (in F)',
     main = 'Max and min temperatures in Boston on June 7th\n2007-2017\n',
     ylim = c(40, 100), 
     col = 'red')
lines(x = boston.weather.min$year, 
      y = boston.weather.min$min.temp,
      col = 'blue')
legend(2014, 100, 
       legend = c('Max', 'Min'),
       col = c('red', 'blue'), lty = 1)

We might also want to create a scatterplot to see if there is a correlation between the lowest and highest temperature for each year. For example, do years with low minimum temperatures also have low maximum temperatures?

Scatter plot

The plot() function by default creates a scatter plot.

plot(x = boston.weather.min$min.temp, 
     y = boston.weather.max$max.temp)

We can draw the straight line that best fits the relationship between these two variables. In order to do this, we can use abline(), a function that adds straight lines to a plot, in combination with an additional argument, lm(), which stands for "linear model." We have to tell R that boston.weather.max$max.temp is our y-variable (aka "outcome" variable) and boston.weather.min$min.temp is our x variable (aka "input" variable) for the regression.

Highlight the following code and run it all at once:

plot(x = boston.weather.min$min.temp, 
     y = boston.weather.max$max.temp)
abline(lm(boston.weather.max$max.temp ~ boston.weather.min$min.temp))

Here's one way to calculate the correlation coefficient between these two variables.

cor(x = boston.weather.min$min.temp, 
     y = boston.weather.max$max.temp)

## [1] 0.7049562

We can also print out the regression output.

summary(lm(boston.weather.max$max.temp ~ boston.weather.min$min.temp))

## 
## Call:
## lm(formula = boston.weather.max$max.temp ~ boston.weather.min$min.temp)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -12.9355  -2.8148   0.2232   2.0267  17.4810 
## 
## Coefficients:
##                             Estimate Std. Error t value Pr(>|t|)  
## (Intercept)                 -27.7097    34.1317  -0.812   0.4378  
## boston.weather.min$min.temp   1.7917     0.6009   2.982   0.0154 *
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 8.171 on 9 degrees of freedom
## Multiple R-squared:  0.497,	Adjusted R-squared:  0.4411 
## F-statistic: 8.891 on 1 and 9 DF,  p-value: 0.0154

Histogram

Now we have a sense of the minimum and maximum Boston temperatures on June 7th of each year, but what if we wanted to get a better sense of the distribution of temperatures?

Let's go back to the hourly data, boston.weather:

head(boston.weather)

## # A tibble: 6 x 10
##   time    temp dewpoint windspeed precip events conditions     year month
##   <time> <dbl>    <dbl>     <dbl>  <dbl> <chr>  <chr>         <int> <int>
## 1 00:54   54       39.9       8.1     NA <NA>   Clear          2007     6
## 2 01:54   53.1     39.9       9.2     NA <NA>   Partly Cloudy  2007     6
## 3 02:54   53.1     41         9.2     NA <NA>   Clear          2007     6
## 4 03:54   52       41         8.1     NA <NA>   Clear          2007     6
## 5 04:54   51.1     41         9.2     NA <NA>   Clear          2007     6
## 6 05:54   52       41         9.2     NA <NA>   Clear          2007     6
## # ... with 1 more variable: day <int>

We could plot a histogram of temp using hist(), which would give a sense of the most common temperatures in the data.

hist(boston.weather$temp)

This is nice--we can see that most of the time on June 7th, hourly temperatures are in the 60 or so degree range.

What if I wanted to see the distribution of temperatures by year?

Boxplot

I really like boxplots. Here is a good article from Khan Academy that reviews what boxplots do: https://www.khanacademy.org/math/statistics-probability/summarizing-quantitative-data/box-whisker-plots/a/box-plot-review. Basically, boxplots provide a summary of the range of values: not only the min and the max, but also the first quartile, the median, and the third quartiles. Here is a piece in the Opinion section of the New York Times, about the cost of college, with an excellent series of boxplots right up top: https://www.nytimes.com/interactive/2018/06/05/opinion/columnists/what-college-really-costs.html.

Let's make boxplots of the hourly temperature for each year.

?boxplot()
boxplot(temp~year, 
        data = boston.weather)

A taste of ggplot2

If we have time, let's use a the built-in Seatbelts data of drivers killed by year in Great Britain to showcase qplot(), a function within the ggplot2 data viz package.

Make the data.seatbelts dataframe:

# Create a data frame 
data(Seatbelts)
data.seatbelts <- data.frame(Year = floor(time(Seatbelts)),
                  Month = factor(cycle(Seatbelts),
                               labels = month.abb), Seatbelts)

How would you inspect the first few rows of data.seatbelts?

qplot() stands for "quick plot" and is the quick and simple plotting function of ggplot2.

# Now, let's plot the data, using the basic plotting function of ggplot2 
qplot(data = data.seatbelts, 
      x = Year, 
      y = DriversKilled, 
      main = 'Drivers Killed by Year')

Don't worry, we'll get more into what the data actually say in the next class. What do you see as some advantages to ggplot2?

What else?

• How would you add axis labels and a title to the boxplot of hourly Boston temperatures?
• What are some positive aspects and limitations to the base R graphics package?
• Are there other chart types or data visualziation methods you are interested in exploring?