sum()

Use summarize() and sum() to find the total credits from all the scrap.

summarize(scrap, total_credits = sum(credits))

mean()

Use summarize() and mean() to calculate the mean price_per_pound in the scrap report.

summarize(scrap, mean_price = mean(price_per_pound, na.rm = T))

median()

Use summarize to calculate the median price_per_pound in the scrap report.

summarize(scrap, median_price = median(price_per_pound, na.rm = T))

max()

Use summarize to calculate the maximum price per pound any scrapper got for their scrap.

summarize(scrap, max_price = max(price_per_pound, na.rm = T))

min()

Use summarize to calculate the minimum price per pound any scrapper got for their scrap.

summarize(scrap, min_price = min(price_per_pound, na.rm = T))

sd()

What is the standard deviation of the credits?

summarize(scrap, stdev_credits = sd(credits))

quantile()

Quantiles are useful for finding the upper or lower range of a column. Use the quantile() function to find the the 5th and 95th quantile of the prices.

summarize(scrap, 
          price_5th_pctile  = quantile(price_per_pound, 0.05, na.rm = T),
          price_95th_pctile = quantile(price_per_pound, 0.95))

Hint: Add na.rm = T to quantile().

n()

n() stands for count.

Use summarize and n() to count the number of reported scrap records.

summarize(scrap, scrap_records = n())

summary <- summarize(scrap, 
                     max_credits      = __________,
                     weight_90th_pct  = quantile(Weight, 0.90),
                     count_records    = __________,

niima_scrap <- filter(scrap, destination == "Niima Outpost") 

niima_scrap <- summarize(niima_scrap, scrap_records = n())

Too much

That sounds like a whole lot of summarizing.

It’d be nice if we could easily find the mean for every city. Then we could summarize once and move on.

Bargain hunters

Who’s selling goods for cheap? Use group_by with the column Origin, and then usesummarize to find the mean(price_per_pound) at each Origin City.

scrap_summary <- group_by(scrap, origin) %>%
                 summarize(mean_price =  mean(price_per_pound, na.rm = T))

scrap_means <- group_by(scrap, origin) %>%
                 summarize(mean_price       =  mean(price_per_pound, na.rm = T),
                           mean_price_round = round(mean_price, digits = 2)) %>%
                 arrange(mean_price_round) %>%
                 ungroup()

Who’s making lots of transactions? Try using group_by with the column origin and then summarize to count the number of scrap records at each city.

scrap_counts <- group_by(scrap, origin) %>%
                summarize(origin_count  =  n()) %>%
                ungroup()

[If this is true], "Do this", "Otherwise do this"

Here’s a handy ifelse statement to help you identify lightsabers.

Or say we want to label all the porgs over 60 cm as tall, and everyone else as short. Whenever we want to add a column where the value depends on the value found in another column. We can use ifelse().

Or maybe we’re trying to save some money and want to flag all the items that cost less than 500 credits. How?

mutate() + ifelse() is powerful!

# Add an affordable column
scrap <- scrap %>% 
         mutate(affordable = ifelse(price_per_unit < 500, 
                                    "Cheap", 
                                    "Expensive"))

Plot the data, Plot the data, Plot the data

The ggplot() sandwich

A ggplot has 3 ingredients.

1. The base plot

library(ggplot2)

ggplot(scrap)

we load version 2 of the package library(ggplot2), but the function to make the plot is only ggplot(). No 2. Sorry.

2. The the X, Y aesthetics

The aesthetics assign the columns from the data that you want to use in the chart. This is where you set the X-Y variables that determine the dimensions of the plot.

ggplot(scrap, aes(x = origin, y = amount)) 

3. The layers or geometries

ggplot(scrap, aes(x = origin, y = amount)) + geom_col()

Colors

Now let’s change the fill color to match the origin.

ggplot(scrap, aes(x = origin, y = amount, fill = origin)) +
  geom_col() 

ggplot(scrap, aes(x = destination, y = amount )) + geom_col()

ggplot(scrap, aes(x = __column1__, y = __column2__)) + geom_point()

Colors

Now let’s use color to show the origins of the scrap

ggplot(scrap, aes(x = destination, y = credits, color = origin)) +
  geom_point()

This is a A LOT of detail. Let’s make a bar chart and add up the sales to make it easier to understand.

ggplot(scrap, aes(x = destination, y = credits, fill = origin)) + geom_col()

It’s still tricky to compare sales by origin. Let’s change the position of the columns.

ggplot(scrap, aes(x = destination, y = credits, fill = origin)) +
  geom_col(position = "dodge")

Colors

Now let’s use color to show the destinations of the scrap.

ggplot(scrap, aes(x = origin, y = credits, color = destination)) +
  geom_point()

Bar charts

This is way too much detail. Let’s simplify and make a bar chart that adds up all the sales. Note that we use fill= inside aes() instead of color=. If we use color, we get a colorful outline and gray bars.

ggplot(scrap, aes(x = origin, y = credits, fill = destination)) +
  geom_col()

Let’s change the position of the bars to make it easier to compare sales by destination for each origin? Remember, you can use help(geom_col) to learn about the different options for that plot. Feel free to do the same with other geom_’s as well.

ggplot(scrap, aes(x = origin, y = credits, fill = destination)) +
  geom_col(position = "dodge")

Facet wrap

Does the chart feel crowded to you? Let’s use the facet wrap function to put each origin on a separate chart.

ggplot(scrap, aes(x = origin, y = credits, fill = destination)) +
  geom_col(position = "dodge") +
  facet_wrap("destination")

Labels

We can add lables to the chart by adding the labs() layer. Let’s give our chart from above a title.

Titles and labels

ggplot(scrap, aes(x = origin, y = credits, fill = destination)) +
  geom_col(position = "dodge") +
  facet_wrap("destination") +
  labs(title    = "Scrap sales by origin and destination",
       subtitle = "Planet Jakku",
       x        = "Origin", 
       y        = "Total sales")

Add lines

More layers! Let’s say we were advised to avoid sales that were over 50 Billion credits. Let’s add that as a horizontal line to our chart. For that, we use geom_hline().

Reference lines

ggplot(scrap, aes(x = origin, y = credits, fill = destination)) +
  geom_col(position = "dodge") +
  facet_wrap("destination") +
  labs(title    = "Scrap sales by origin and destination",
       subtitle = "Planet Jakku",
       x        = "Origin", 
       y        = "Total sales") +
     geom_hline(yintercept = 5E+10, color = "orange", size = 2)

Drop 2.2e+06 scientific notation

Want to get rid of that ugly scientific notation? We can use options(scipen = 999). Note that this is a general setting in R. Once you use options(scipen = 999) in your current session, you don’t have to use it again. (Like loading a package, you only need to run the line once when you start a new R session).

options(scipen = 999)

ggplot(scrap, aes(x = origin, y = credits, fill = destination)) +
  geom_col(position = "dodge") +
  facet_wrap("destination") +
  theme_bw() +
  labs(title = "Scrap sales by origin and destination", 
       x     = "Origin", 
       y     = "Total sales")

Themes

You may not like the appearance of these plots. ggplot2 uses theme functions to change the appearance of a plot. Try some.

ggplot(scrap, aes(x = origin, y = credits, fill = destination)) +
  geom_col(position = "dodge") +
  facet_wrap("destination") +
  theme_bw()

Save plots

You’ve made some plots you can be proud of, so let’s learn to save them so we can cherish them forever. There’s a function called ggsave to do just that. How do we ggsave our plots?

Let’s try help(ggsave) or ?ggsave.

# Get help
help(ggsave)
?ggsave

# Run the R code for your favorite plot first
ggplot(data, aes()) +
   .... +
   ....


# Then save your plot to a png file of your choosing
ggsave("results/plot_name.png")

# Name the ggplot you want to save
my_plot <- ggplot(...) + geom_point(...)
  
# Save it
ggsave(filename = "results/Save_it_here.png", 
       plot = my_plot)

Learn more about saving plots: http://stat545.com/