Core tidyverse

• ggplot2: ggplot2 is a system for declaratively creating graphics, based on The Grammar of Graphics.
• tibble: tibble is a modern re-imagining of the data frame, keeping what time has proven to be effective, and throwing out what it has not.
• tidyr: tidyr provides a set of functions that help you get to tidy data.
• readr: readr provides a fast and friendly way to read rectangular data (like csv).
• purrr: purrr provides a complete and consistent set of tools for working with functions and vectors.
• dplyr: dplyr provides a grammar of data manipulation

read.csv() vs. read_csv()

First, let’s use the usual way to read csv data into R (make sure that you are using the right working directory).

df_original <- read.csv(file="HousePrices.csv", 
                        header=TRUE, stringsAsFactors=FALSE)
class(df_original) # this is a data frame

## [1] "data.frame"

We can also use read_csv() from the readr package in the tidyverse to read the same csv file:

df <- read_csv(file="HousePrices.csv")
class(df) # this is a tibble (tbl_df)

## [1] "spec_tbl_df" "tbl_df"      "tbl"         "data.frame"

Your turn

If you haven’t done this already (from lab-3), download HousePrices.csv and put it into your working directory.

Run the following code:

install.packages("tidyverse")
library(tidyverse)
df_original <- read.csv(file="HousePrices.csv", 
                        header=TRUE, stringsAsFactors=FALSE)
df <- read_csv(file="HousePrices.csv")

And then print these two data sets by entering their names to the console. What are your observations?

df_original
df

Package dplyr

dplyr provides the following verbs for data manipulation.

1. select
2. filter
3. arrange
4. mutate
5. summarise & group_by
6. joining (merging) data frames / tables

We will be using HousePrices.csv throughout this lab. Please download and put the file in your working directory. (Use getwd and setwd to make sure you have a correct working directory.)

select(): Pick columns by name

select(df, c(price, lot_size)) # equivalent to df[, c("price", "lotsize")]

## # A tibble: 1,728 x 2
##     price lot_size
##     <dbl>    <dbl>
##  1 132500     0.09
##  2 181115     0.92
##  3 109000     0.19
##  4 155000     0.41
##  5  86060     0.11
##  6 120000     0.68
##  7 153000     0.4 
##  8 170000     1.21
##  9  90000     0.83
## 10 122900     1.94
## # … with 1,718 more rows

Chaining/Pipelining

Pipe operator %>% makes the code much readable. Essentially, you send the data through a set of operations and the operations are connected by a pipe.

Note: You must have a space before and after %>%

# in tidyverse, we use dot/period to represent whatever comes from the previous stage
df %>% select(., c(price, lot_size)) # notice that df is outside

## # A tibble: 1,728 x 2
##     price lot_size
##     <dbl>    <dbl>
##  1 132500     0.09
##  2 181115     0.92
##  3 109000     0.19
##  4 155000     0.41
##  5  86060     0.11
##  6 120000     0.68
##  7 153000     0.4 
##  8 170000     1.21
##  9  90000     0.83
## 10 122900     1.94
## # … with 1,718 more rows

# if the first argument is a dot/period, you may omit it in your code
# the packages in the tidyverse will automatically fill it in for you
df %>% select(c(price, lot_size))

## # A tibble: 1,728 x 2
##     price lot_size
##     <dbl>    <dbl>
##  1 132500     0.09
##  2 181115     0.92
##  3 109000     0.19
##  4 155000     0.41
##  5  86060     0.11
##  6 120000     0.68
##  7 153000     0.4 
##  8 170000     1.21
##  9  90000     0.83
## 10 122900     1.94
## # … with 1,718 more rows

df %>% select(price:heat) # all columns between price and fuel

## # A tibble: 1,728 x 9
##     price lot_size waterfront   age land_value construction air_cond fuel  heat 
##     <dbl>    <dbl> <chr>      <dbl>      <dbl> <chr>        <chr>    <chr> <chr>
##  1 132500     0.09 No            42      50000 No           No       Elec… Elec…
##  2 181115     0.92 No             0      22300 No           No       Gas   Hot …
##  3 109000     0.19 No           133       7300 No           No       Gas   Hot …
##  4 155000     0.41 No            13      18700 No           No       Gas   Hot …
##  5  86060     0.11 No             0      15000 Yes          Yes      Gas   Hot …
##  6 120000     0.68 No            31      14000 No           No       Gas   Hot …
##  7 153000     0.4  No            33      23300 No           No       Oil   Hot …
##  8 170000     1.21 No            23      14600 No           No       Oil   Hot …
##  9  90000     0.83 No            36      22200 No           No       Elec… Elec…
## 10 122900     1.94 No             4      21200 No           No       Gas   Hot …
## # … with 1,718 more rows

# select columns that contain "room" in their column name
df %>% select(contains("room")) 

## # A tibble: 1,728 x 2
##    bathrooms rooms
##        <dbl> <dbl>
##  1       1       5
##  2       2.5     6
##  3       1       8
##  4       1.5     5
##  5       1       3
##  6       1       8
##  7       1.5     8
##  8       1.5     9
##  9       1.5     8
## 10       1.5     6
## # … with 1,718 more rows

Your turn

Select columns age and land_value from df

filter(): Keep rows that match criteria

df %>% filter(price < 200000, fuel == "Electric")

## # A tibble: 246 x 15
##       X1  price lot_size waterfront   age land_value construction air_cond fuel 
##    <dbl>  <dbl>    <dbl> <chr>      <dbl>      <dbl> <chr>        <chr>    <chr>
##  1     1 132500     0.09 No            42      50000 No           No       Elec…
##  2     9  90000     0.83 No            36      22200 No           No       Elec…
##  3    13  85860     8.97 No            13       4800 No           No       Elec…
##  4    21 112000     1    No            12       8600 No           No       Elec…
##  5    22 104900     0.43 No            21       5600 No           No       Elec…
##  6    25  90400     0.36 No            16       5200 No           No       Elec…
##  7    60 175000     0.47 No            15      27200 No           No       Elec…
##  8    65 171000     3.16 No            15      24100 No           No       Elec…
##  9    69 182000     1    No            16      26600 No           No       Elec…
## 10    81 114000     0.06 No            14       1600 No           No       Elec…
## # … with 236 more rows, and 6 more variables: heat <chr>, sewer <chr>,
## #   living_area <dbl>, fireplaces <dbl>, bathrooms <dbl>, rooms <dbl>

Chaining multiple operations

df %>% 
  filter(price < 200000, fuel == "Electric") %>%
  select(price, fuel, air_cond)

## # A tibble: 246 x 3
##     price fuel     air_cond
##     <dbl> <chr>    <chr>   
##  1 132500 Electric No      
##  2  90000 Electric No      
##  3  85860 Electric No      
##  4 112000 Electric No      
##  5 104900 Electric No      
##  6  90400 Electric No      
##  7 175000 Electric No      
##  8 171000 Electric No      
##  9 182000 Electric No      
## 10 114000 Electric No      
## # … with 236 more rows

Note: In the interest of readability, put different operations in separate lines.

After the operations, you may want to save the result as a new data frame. You can then use the new data frame for other analyses.

new_df <- df %>% 
  filter(price < 200000, fuel == "Electric") %>%
  select(price, fuel, air_cond)

nrow(new_df)

## [1] 246

Your turn

Use pipe %>% to chain the following two operations

1. Select columns stories and driveway from df
2. Include only houses that have no driveway

arrange(): Reorder rows

df %>%
  select(price, air_cond, rooms) %>%
  arrange(price)

## # A tibble: 1,728 x 3
##    price air_cond rooms
##    <dbl> <chr>    <dbl>
##  1  5000 Yes          6
##  2 10300 No           4
##  3 10300 No           6
##  4 20000 No           4
##  5 25000 No           6
##  6 45000 No           4
##  7 49387 No           5
##  8 58500 No           5
##  9 60000 No           6
## 10 60000 No           7
## # … with 1,718 more rows

use desc() for a descending order

df %>%
  select(price, air_cond, rooms) %>%
  arrange(desc(price))

## # A tibble: 1,728 x 3
##     price air_cond rooms
##     <dbl> <chr>    <dbl>
##  1 775000 Yes          9
##  2 775000 Yes         12
##  3 760000 Yes         12
##  4 725000 Yes         10
##  5 670000 No           8
##  6 658300 Yes         11
##  7 655000 No           6
##  8 650000 Yes         10
##  9 650000 Yes         12
## 10 649000 Yes         12
## # … with 1,718 more rows

Your turn

1. Select houses with Private sewer
2. Order the results in a descending order of lot_size

mutate(): Add new variables

Create new variables that are functions of existing variables

df %>%
  mutate(bedrooms = rooms - bathrooms) %>%
  filter(bedrooms>3) %>%
  select(bedrooms, bathrooms, rooms)

## # A tibble: 1,381 x 3
##    bedrooms bathrooms rooms
##       <dbl>     <dbl> <dbl>
##  1      4         1       5
##  2      3.5       2.5     6
##  3      7         1       8
##  4      3.5       1.5     5
##  5      7         1       8
##  6      6.5       1.5     8
##  7      7.5       1.5     9
##  8      6.5       1.5     8
##  9      4.5       1.5     6
## 10     11         1      12
## # … with 1,371 more rows

Your turn

• Use mutate to create a new column, expensive_house, which equals “yes” if the house price is greater than 150000 and “no” otherwise.
  • Hint: Recall the ifelse() function
• After that, select the price and expensive_house columns and arrange the rows in the descending order of price.

## # A tibble: 1,728 x 2
##     price expensive_house
##     <dbl> <chr>          
##  1 775000 yes            
##  2 775000 yes            
##  3 760000 yes            
##  4 725000 yes            
##  5 670000 yes            
##  6 658300 yes            
##  7 655000 yes            
##  8 650000 yes            
##  9 650000 yes            
## 10 649000 yes            
## # … with 1,718 more rows

summarise(): Reduce variables to values

• group_by() creates the groups that will be operated on
• summarise() uses the provided aggregation function to summarise each group

df %>%
  group_by(air_cond) %>%
  summarise(avg_price = mean(price))

## # A tibble: 2 x 2
##   air_cond avg_price
##   <chr>        <dbl>
## 1 No         187022.
## 2 Yes        254904.

You can have multiple summary/aggregate statistics:

df %>%
  group_by(air_cond) %>%
  summarise(n_house = n(), # n() gives the number of rows in each group
            avg_price = mean(price))  

## # A tibble: 2 x 3
##   air_cond n_house avg_price
##   <chr>      <int>     <dbl>
## 1 No          1093   187022.
## 2 Yes          635   254904.

Your turn

Some houses have a drive way, some don’t. Use group_by and summarise to obtain the following:

## # A tibble: 2 x 4
##   construction n_house min_lotsize max_lotsize
##   <chr>          <int>       <dbl>       <dbl>
## 1 No              1647        0          12.2 
## 2 Yes               81        0.01        3.47

Joining two data frames (or tables)

Note: left_join(a, b, by="x1") is equivalent to a %>% left_join(b, by="x1")

Lab assignment

This lab assignment involves 2 tasks (see the following slides). Once you finish the following tasks, please put everything in one single R file with the file name assignment2.R (.R is the file extension) and upload it to iCollege (Lab Assignment 2).

You will lose 50% of the points if you use a different file name or put your code in multiple files.

In addition, lab assignments will be graded based on:

• Accuracy: whether the R script achieves the objectives
• Readability: whether the R script is clean, well-formatted, and easily readable
  • You risk loosing 10 points if your code has no proper indentation or has more than 80 characters in a line.

Lab Assignment 1/2

1.1. Show price, air_cond, heat, fireplaces for houses that have no fireplaces

1.2. Create a new variable price_per_bedroom which is price divided by the number of bedrooms. Show only price, bedrooms, and price_per_bedroom columns and arrange the rows in the descending order of price_per_bedroom

1.3. Create a new variable has_4_or_more_bedrooms which is TRUE if the house has 4 or more bedrooms and FALSE otherwise. Use this variable and group_by and summarise() to find how many houses have 4 or more bedrooms and how many don’t