1 Good morning, Detectives!

Please open your RStudio project from Day 1

1. Open your project folder from Day 1.
2. Double click the .Rproj file to open RStudio.
3. Relax.

library("readr")
library("dplyr")

# Your code starts here.

all_cats <- read_csv("data/missing_cat_list.csv")

# Sort from low to high
arrange(all_cats, age)

# Sort from high to low
arrange(all_cats, desc(age))

# Filter to cats 5 years old
cats_5yo <- filter(all_cats, age == 5)

# Filter to cats from UK or Australia
cricket_cats <- filter(all_cats, country %in% c("UK", "Australia"))

2 | You have Voicemail (1)

Put your detective hat back on. Your neighbor just called and he shared some bad news. His cat has gone missing. He was so worked-up the only thing you could make out in the message was a description of how great his cat’s personality was. Here are the few details you were able to jot down.

My neighbor’s great cat:

The cat’s clumsiness and its greediness sum to more than 12.

You could barely make this out, but you think the cat’s grumpiness and playfulness only sum to 6.

And it’s a female cat.

Oh and the cat is the same age as your neighbor (in human years), who is probably around 85, maybe younger.

It sounds like we’re going to need some math for this puzzle. Let’s calculate some new columns with mutate() to help find your neighbor’s cat.

3 | mutate()

It’s often useful to edit existing columns in a data frame or add new columns that are functions of existing columns. That’s the job of mutate().

all_cats <- mutate(all_cats, 
                   human_age = age * 4 + 20)

my_cat <- mutate(my_cat, 
                 found_by = "Pet Detective Cooper")

my_cat <- mutate(my_cat, 
                 found_by = "Nevermind, it's a secret")

3.3 Find your neighbor’s cat

Now let’s see if you can use mutate() to help match the description of the cat’s personality.

Here’s a reminder of how your neighbor described his cat.

The cat’s clumsiness and its greediness scores sum to more than 12. You could barely make this out, but you think the cat’s grumpiness and playfulness only sum to 6. It’s a female cat.

And the cat is the same age as him (in human years), which is probably around 85, maybe less.

Use mutate() to add columns to the missing cat list, and then filter() the new columns to narrow the list down to your neighbor’s cat.

Here’s a snippet to get you started.

library("dplyr")

# Add columns to all_cats
all_cats <- mutate(all_cats, 
                  clumsy_and_greedy  = clumsy + greedy,
                  grumpy_and_playful = grumpy + ...,
                  human_age = age * 4 + ...
                  )

# Filter for cats matching neighbor's description
filter(all_cats,
         clumsy_and_greedy  > ...,
         grumpy_and_playful == ...,
         human_age > ...
         )

You can chain these two functions together and do everything in one go. For that you can use the %>% (pipe).

In a script the %>% is read as “and then”. In the code below we are telling R to mutate the data and then to filter it.

nei_cat <- mutate(all_cats, 
                  clumsy_and_greedy  = clumsy + greedy,
                  grumpy_and_playful = grumpy + playful,
                  human_age = age * 4 + 20) %>%
           filter(clumsy_and_greedy  > 12,
                  grumpy_and_playful == 6,
                  human_age > 80)

Pop Quiz, hotshot!

What is the name of your friend’s cat?

Damon
Fluffy George
Stinkerbell
Precious Abe

Show solution

Fluffy George

Breaking Meows! You rock!

4 | The case of the baby tigers 🐯 🐯

Sitting down for your morning coffee you notice the front page of the paper has a story on cats!
According to Phyllis Cattleby there’s been a whole string of striped kittens gone missing. This sounds like a case for a pet detective.

Phyllis is pointing her finger at a circus that just came to town. A circus that seems to be drawing big crowds with their adorable and surprisingly tame “tiger babies”.

What evidence could we find in the cat database to support or refute the claim that the circus is stealing young striped cats?

For this level of sleuthing we’re going to need to summarize our data.

5 | summarize() this

summarize allows you to apply a summary function like median() to a column and collapse your data down to a single row. To really dig into summarize you’ll want to know some common summary functions, such as sum(), mean(), median(), min(), and max().

summarize(all_cats, total_greedy = sum(greedy))

cat_summary <- summarize(all_cats, 
                         mean_age = mean(age, na.rm = T))

summarize(all_cats, 
          age_5th_pctile  = quantile(age, 0.05, na.rm = T),
          age_95th_pctile = quantile(age, 0.95))

summarize(all_cats, median_age = median(age, na.rm = T))

filter(all_cats, color == "striped") %>% summarize(striped_med_age = median(age, na.rm = T))

6 | group_by()

Enter group_by() stage left. If you thought summarize was awesome, wait until you include group_by with your summarize commands.

Try using group_by with the column color and then use summarize to count the number of cats in each group.

group_by(all_cats, color) %>% summarize(color_count  =  n()) %>% ungroup()

group_by(all_cats, color) %>% 
      summarize(mean_age =  mean(age, na.rm = T)) %>% ungroup()

group_by(all_cats, color) %>% 
  summarize(mean_age       =  mean(age, na.rm = T),
            mean_age_round = round(mean_age, digits = 1)) %>%  
  arrange(mean_age_round) %>% ungroup()

7 | Save files

Let’s save the last summary table we created to a CSV. That way we can print it to have it faxed to the police later. To save a data frame we’ll use the write_csv() function from our favorite readr package.

# First give the new data a name
ages_by_color <- group_by(all_cats, color) %>% 
                    summarize(mean_age       =  mean(age, na.rm = T),
                              mean_age_round = round(mean_age, digits = 1)) %>%  
                    arrange(mean_age_round) %>% ungroup()

# Write the file to your project folder
write_csv(ages_by_color, "mean_cat_ages_by_color.csv")

8 | Grouped mutate()

We can bring back mutate to add a column based on the grouped values in a data set. For example, you may want to add a column showing the average age by country to the whole table.

When you combine group_by and mutate the new column will be calculated based on the values within each group.

group_by(all_cats, country) %>% mutate(country_mean_age = mean(age, na.rm = T)) %>% ungroup()

group_by(all_cats, country) %>% 
  mutate(grumpy_by_country = median(grumpy, na.rm = T)) %>% 
  ungroup()

8.1 Break time

Take 5 minutes to relax.

9 Plots with ggplot2

Install ggplot2 using install.packages("ggplot2").

library("readr")
library("dplyr")
library("ggplot2")

# Path to movie data
movie_url <- "https://raw.githubusercontent.com/MPCA-air/RCamp/master/data/movies/IMDB.csv"


# Read the IMDB movie data and save as `movies`
movies <- read_csv(movie_url)

# Show column names
#names(movies)

# Rename the 'actor_1_name' column
movies <- rename(movies, superstar = actor_1_name)

ggplot(movies)

ggplot(movies, aes(x = movie_facebook_likes, y = gross_mil)) 

ggplot(movies, aes(x = movie_facebook_likes, y = gross_mil)) + 
  geom_point()

# This will work
ggplot() +
  geom_point()

# BUT this will give you a nasty error message
ggplot()
  + geom_point()

ggplot(movies, aes(x = ?column1, y = ?column2)) + geom_point()

9.3 Scatterplots and point charts.

Let’s select only recent movies using filter().

filter(movies, title_year >= 2010)

9.3.1 To repeat the earlier chart with the filtered data use the %>% pipe.

filter(movies, title_year >= 2010) %>%
  ggplot(aes(x = movie_facebook_likes, y = gross_mil)) + 
    geom_point()

9.3.2 Add transparency to the points.

filter(movies, title_year >= 2010) %>%
  ggplot(aes(x = movie_facebook_likes, y = gross_mil)) + 
    geom_point(alpha = 0.1) 

9.3.3 Add a regression line.

We can keep adding layers! You can build your plot sandwich as big as you like.

Use geom_smooth() to add a regression line.

filter(movies, title_year >= 2010) %>% 
  ggplot(aes(x = movie_facebook_likes, y = gross_mil)) + 
    geom_point(alpha = 0.25) +
    geom_smooth(method = "lm")

Exercise

Make a scatterplot of imdb_score and gross_mil with a fitted line showing the relationship.

Show solution

Stop cheating! Just kidding here’s some code to help.

filter(movies, title_year >= 2010) %>% 
  ggplot(aes(x = imdb_score, y = gross_mil)) + 
    geom_point(alpha = 0.25) +
    geom_smooth(method = "lm")

ggplot(movies, aes(x = title_year)) + geom_histogram()

ggplot(movies, aes(x = title_year)) + geom_histogram(binwidth = 10)

ggplot(movies, aes(x = title_year, fill = movie_color)) +
  geom_histogram(binwidth = 10)

ggplot(movies, aes(x = title_year, fill = movie_color)) +
  geom_histogram(binwidth = 10, position = "dodge")

ggplot(movies, aes(x = title_year, fill = movie_color)) +
  geom_histogram(binwidth = 10) +
  facet_wrap(~ movie_color)

ggplot(movies, aes(title_year, fill = movie_color)) +
  geom_histogram(binwidth = 10) +
  facet_wrap(~ movie_color, scales = "free_y")

# Stacked
ggplot(movies, aes(title_year, fill = content_rating)) +
  geom_histogram(binwidth = 10) 

# Side by side
ggplot(movies, aes(title_year, fill = content_rating)) +
  geom_histogram(binwidth = 10, position = position_dodge()) 

# Separate charts
ggplot(movies, aes(title_year, fill = content_rating)) +
  geom_histogram(binwidth = 10) + 
  facet_wrap( ~ content_rating)

# Free scale it
ggplot(movies, aes(title_year, fill = content_rating)) +
  geom_histogram(binwidth = 10) + 
  facet_wrap( ~ content_rating, scales = "free_y")

10 Day 2 homework

1. Load some data into R from a recent project of yours.
   • If you have data in Excel, you can export a .CSV file by choosing “Save as” and selecting CSV from the file type menu.
2. Create 2 types of plots using the data.
3. Paste the function runif(1) into your console.
   • If the number you get is bigger than 0.2, email your charts to the rest of the class.