Working With Data Frames

data science data types programming R

Working with data frames in R.

Danielle Brantley https://gist.github.com/danielle-b
01-30-2020

Basketball has been heavy on mind lately. One reason being the All the Smoke podcast that I listen to with Matt Barnes and Stephen Jackson every week. I’ve been listening to the podcast since day one. Another reason has been the recent passing of Kobe Bryant, his daughter Gianna and the seven other lives that were taken in that helicopter accident. When I think of Kobe, I think about my brother and the kids I went to school with. They would shout ‘KOBE’ anytime they threw something.

Because of people like my brother, like those kids I went to school with, I decided to go with a basketball theme with this post. I wanted to shed light on some of the players that Kobe inspired and some that maybe inspired him. The data set that I will use for this blog post was posted on Kaggle by Jason Baruch. Baruch created the NBA Player of the Week data set that analyzes player of the week data from 1979-80 season to the present. I decided to download the csv and play around a bit. With that said, let’s get into data frames in R.

Data frames are a list of vectors of equal length. They can contain multiple data types and they are probably the most common data structure in R. Not only did I learn about data frames but I learned about some R packages that makes doing data science with data frames efficient.

Installing and Loading Packages

The first thing I did as part of the DataQuest lesson was to install the readr package, which is part of the tidyverse. Readr is designed to improve data science workflow. DataQuest gives me a few ways that readr does this:

After installing readr, I loaded the package using the library function.

Importing Data with R

RStudio (the IDE I’ve been using in my posts) has an option to import data. To do this I went to File -> Import Dataset -> From Text(readr). This leads to a screen that prompts me type a file name or URL. However, I decided to click on Browse which leads me to import a document. When I imported my csv that I downloaded from Kaggle, this is what I get:

After I click import, my csv file is opened like this.

Note: The tidyverse is a collection of packages designed to make using R for data science more effective. In addition to readr, tidyverse includes dplyr which will be discussed later. Instead of installing each package, I installed tidyverse just as I did with readr example in the above screenshots.

library(tidyverse)

NBA_player_of_the_week <- read_csv('/Users/User/datascidani2/_posts/working_with_dataframes 01-30-20/NBA_player_of_the_week.csv')
View(NBA_player_of_the_week)

Tibbles

To make the data easier to work with, I decided to get my data frame into a tibble. A tibble is the same as a data frame in base R. DataQuest mentions that they are two-dimensional data structures that store data of multiple types. They also have very important advantages: clarity, consistency and printing. You can type the name of a tibble and R will only print the first 10 rows.

tibble::as_tibble(NBA_player_of_the_week)
# A tibble: 1,334 x 17
   Player Team  Conference Date  Position Height Weight   Age
   <chr>  <chr> <chr>      <chr> <chr>    <chr>   <dbl> <dbl>
 1 Ben S… Phil… East       20-J… PF       6'10      230    23
 2 Kawhi… Los … West       20-J… F        6'7       230    28
 3 Josh … Phil… East       13-J… G        6'6       200    26
 4 DeMar… San … West       13-J… GF       6'7       220    30
 5 Giann… Milw… East       6-Ja… F        6'11      242    25
 6 LeBro… Los … West       6-Ja… F        6'8       250    35
 7 Jayle… Bost… East       30-D… SF       6'7       220    23
 8 Brand… New … West       30-D… SF       6'9       190    22
 9 Kyle … Toro… East       23-D… PG       6'1       196    33
10 Denni… Okla… West       23-D… PG       6'1       172    26
# … with 1,324 more rows, and 9 more variables: Draft_Year <dbl>,
#   Seasons_in_league <dbl>, Season <chr>, Season_short <dbl>,
#   Pre_draft_Team <chr>, Real_value <dbl>, Height_CM <dbl>,
#   Weight_KG <dbl>, Last_Season <dbl>

Indexing Data Frames

For data frames, I can use indexing to return a specific, row, column or value. It’s pretty similar to indexing matrices. If I wanted to index the column “Team”, I could do this in a couple of ways:

By position :

NBA_player_of_the_week[, 2]
# A tibble: 1,334 x 1
   Team                 
   <chr>                
 1 Philadelphia Sixers  
 2 Los Angeles Clippers 
 3 Philadelphia Sixers  
 4 San Antonio Spurs    
 5 Milwaukee Bucks      
 6 Los Angeles Lakers   
 7 Boston Celtics       
 8 New Orleans Pelicans 
 9 Toronto Raptors      
10 Oklahoma City Thunder
# … with 1,324 more rows

By column name:

NBA_player_of_the_week[, "Team"]
# A tibble: 1,334 x 1
   Team                 
   <chr>                
 1 Philadelphia Sixers  
 2 Los Angeles Clippers 
 3 Philadelphia Sixers  
 4 San Antonio Spurs    
 5 Milwaukee Bucks      
 6 Los Angeles Lakers   
 7 Boston Celtics       
 8 New Orleans Pelicans 
 9 Toronto Raptors      
10 Oklahoma City Thunder
# … with 1,324 more rows

DataQuest also mentioned that I could use a $ symbol to specify a column though when I tried this method R printed 1000 rows of data.

Here are some more examples of indexing data frames:

NBA_player_of_the_week$Team[2]
[1] "Los Angeles Clippers"
NBA_player_of_the_week[6,2]
# A tibble: 1 x 1
  Team              
  <chr>             
1 Los Angeles Lakers
NBA_player_of_the_week[6, ]
# A tibble: 1 x 17
  Player Team  Conference Date  Position Height Weight   Age
  <chr>  <chr> <chr>      <chr> <chr>    <chr>   <dbl> <dbl>
1 LeBro… Los … West       6-Ja… F        6'8       250    35
# … with 9 more variables: Draft_Year <dbl>, Seasons_in_league <dbl>,
#   Season <chr>, Season_short <dbl>, Pre_draft_Team <chr>,
#   Real_value <dbl>, Height_CM <dbl>, Weight_KG <dbl>,
#   Last_Season <dbl>
NBA_player_of_the_week[c(2:8), c("Player", "Pre_draft_Team")]
# A tibble: 7 x 2
  Player                Pre_draft_Team                         
  <chr>                 <chr>                                  
1 Kawhi Leonard         San Diego State                        
2 Josh Richardson       Tennessee                              
3 DeMar DeRozan         USC                                    
4 Giannis Antetokounmpo Filathlitikos Div II Greece (Greece)   
5 LeBron James          St. Vincent St. Mary High School (Ohio)
6 Jaylen Brown          California                             
7 Brandon Ingram        Duke                                   

Working With Data Frame Columns

I could simplify my data frame to contain only variables(the column names) that are relevant for my data analysis.

For example, I want to create a new data frame that includes the names of the players, their teams, characteristics(height, weight, age), draft year and how many seasons they’ve played in the league. I would use dplyr, a tidyverse package designed for analyzing data in data frames.

To use this package, I would install and load it just as I did with the readr package. The dplyr function select() allows me to create a new data frame with only the columns containing the variables I want to use for my analysis. I realized that I already have tidyverse installed so I loaded that before creating this new data frame.

NBA_player_of_the_week_select <- NBA_player_of_the_week %>% select(Player, Team, Position, Height, Weight, Age, Draft_Year, Seasons_in_league)

NBA_player_of_the_week_select
# A tibble: 1,334 x 8
   Player Team  Position Height Weight   Age Draft_Year
   <chr>  <chr> <chr>    <chr>   <dbl> <dbl>      <dbl>
 1 Ben S… Phil… PF       6'10      230    23       2016
 2 Kawhi… Los … F        6'7       230    28       2011
 3 Josh … Phil… G        6'6       200    26       2015
 4 DeMar… San … GF       6'7       220    30       2009
 5 Giann… Milw… F        6'11      242    25       2013
 6 LeBro… Los … F        6'8       250    35       2003
 7 Jayle… Bost… SF       6'7       220    23       2016
 8 Brand… New … SF       6'9       190    22       2016
 9 Kyle … Toro… PG       6'1       196    33       2006
10 Denni… Okla… PG       6'1       172    26       2013
# … with 1,324 more rows, and 1 more variable:
#   Seasons_in_league <dbl>

I could also create new variables and add them to my data frame as columns using another dplyr function called mutate().

NBA_player_of_the_week_mutate <- NBA_player_of_the_week %>% mutate(Draft_Age = Age-(2020-Draft_Year))

NBA_player_of_the_week_mutate
# A tibble: 1,334 x 18
   Player Team  Conference Date  Position Height Weight   Age
   <chr>  <chr> <chr>      <chr> <chr>    <chr>   <dbl> <dbl>
 1 Ben S… Phil… East       20-J… PF       6'10      230    23
 2 Kawhi… Los … West       20-J… F        6'7       230    28
 3 Josh … Phil… East       13-J… G        6'6       200    26
 4 DeMar… San … West       13-J… GF       6'7       220    30
 5 Giann… Milw… East       6-Ja… F        6'11      242    25
 6 LeBro… Los … West       6-Ja… F        6'8       250    35
 7 Jayle… Bost… East       30-D… SF       6'7       220    23
 8 Brand… New … West       30-D… SF       6'9       190    22
 9 Kyle … Toro… East       23-D… PG       6'1       196    33
10 Denni… Okla… West       23-D… PG       6'1       172    26
# … with 1,324 more rows, and 10 more variables: Draft_Year <dbl>,
#   Seasons_in_league <dbl>, Season <chr>, Season_short <dbl>,
#   Pre_draft_Team <chr>, Real_value <dbl>, Height_CM <dbl>,
#   Weight_KG <dbl>, Last_Season <dbl>, Draft_Age <dbl>

As you can see, a new column called Draft_Age was added to my table.

Filtering A Data Frame

Using the filter() function, I could filter a data frame either using a single condition or multiple conditions. This is where the comparison operators come in.

Single Conditions

I could simplify my data frame further by filtering. I would use another dplyr function, filter() to specify conditions that variables must meet in order to be retained in my data frame.

If I want to retain data on players who play for the Eastern Conference, I would type this:

NBA_player_of_the_week_east <- NBA_player_of_the_week %>% filter(Conference == "East")

NBA_player_of_the_week_east
# A tibble: 417 x 17
   Player Team  Conference Date  Position Height Weight   Age
   <chr>  <chr> <chr>      <chr> <chr>    <chr>   <dbl> <dbl>
 1 Ben S… Phil… East       20-J… PF       6'10      230    23
 2 Josh … Phil… East       13-J… G        6'6       200    26
 3 Giann… Milw… East       6-Ja… F        6'11      242    25
 4 Jayle… Bost… East       30-D… SF       6'7       220    23
 5 Kyle … Toro… East       23-D… PG       6'1       196    33
 6 Bam A… Miam… East       16-D… C        6'10      255    22
 7 Jimmy… Miam… East       9-De… GF       6'8       232    30
 8 Giann… Milw… East       2-De… F        6'11      242    25
 9 Spenc… Broo… East       25-N… PG       6'6       210    26
10 Nikol… Orla… East       18-N… PF       7'0       260    29
# … with 407 more rows, and 9 more variables: Draft_Year <dbl>,
#   Seasons_in_league <dbl>, Season <chr>, Season_short <dbl>,
#   Pre_draft_Team <chr>, Real_value <dbl>, Height_CM <dbl>,
#   Weight_KG <dbl>, Last_Season <dbl>

Let’s say I wanted to retain data on players whose weight is greater than 200 pounds. I would write this:

NBA_player_of_the_week_weight <- NBA_player_of_the_week %>% filter(Weight > 200)

NBA_player_of_the_week_weight
# A tibble: 1,004 x 17
   Player Team  Conference Date  Position Height Weight   Age
   <chr>  <chr> <chr>      <chr> <chr>    <chr>   <dbl> <dbl>
 1 Ben S… Phil… East       20-J… PF       6'10      230    23
 2 Kawhi… Los … West       20-J… F        6'7       230    28
 3 DeMar… San … West       13-J… GF       6'7       220    30
 4 Giann… Milw… East       6-Ja… F        6'11      242    25
 5 LeBro… Los … West       6-Ja… F        6'8       250    35
 6 Jayle… Bost… East       30-D… SF       6'7       220    23
 7 Bam A… Miam… East       16-D… C        6'10      255    22
 8 LeBro… Los … West       16-D… F        6'8       250    35
 9 Jimmy… Miam… East       9-De… GF       6'8       232    30
10 Antho… Los … West       9-De… PF       6'10      253    26
# … with 994 more rows, and 9 more variables: Draft_Year <dbl>,
#   Seasons_in_league <dbl>, Season <chr>, Season_short <dbl>,
#   Pre_draft_Team <chr>, Real_value <dbl>, Height_CM <dbl>,
#   Weight_KG <dbl>, Last_Season <dbl>

Multiple Conditions

I was introduced to two new operators: The & operator specifies that both criteria in an expression must be met. The | operator specifies that at least one of the criteria in the expression must be met.

NBA_player_of_the_week_stature <- NBA_player_of_the_week %>% filter(Weight > 200 & Height > "6'5")

NBA_player_of_the_week_stature
# A tibble: 684 x 17
   Player Team  Conference Date  Position Height Weight   Age
   <chr>  <chr> <chr>      <chr> <chr>    <chr>   <dbl> <dbl>
 1 Kawhi… Los … West       20-J… F        6'7       230    28
 2 DeMar… San … West       13-J… GF       6'7       220    30
 3 LeBro… Los … West       6-Ja… F        6'8       250    35
 4 Jayle… Bost… East       30-D… SF       6'7       220    23
 5 LeBro… Los … West       16-D… F        6'8       250    35
 6 Jimmy… Miam… East       9-De… GF       6'8       232    30
 7 Carme… Port… West       2-De… F        6'8       240    35
 8 Spenc… Broo… East       25-N… PG       6'6       210    26
 9 Luka … Dall… West       25-N… SF       6'7       218    21
10 Nikol… Orla… East       18-N… PF       7'0       260    29
# … with 674 more rows, and 9 more variables: Draft_Year <dbl>,
#   Seasons_in_league <dbl>, Season <chr>, Season_short <dbl>,
#   Pre_draft_Team <chr>, Real_value <dbl>, Height_CM <dbl>,
#   Weight_KG <dbl>, Last_Season <dbl>
NBA_player_of_the_week_draft <- NBA_player_of_the_week %>% filter(Draft_Year > 2014 | Seasons_in_league <= 5)

NBA_player_of_the_week_draft
# A tibble: 692 x 17
   Player Team  Conference Date  Position Height Weight   Age
   <chr>  <chr> <chr>      <chr> <chr>    <chr>   <dbl> <dbl>
 1 Ben S… Phil… East       20-J… PF       6'10      230    23
 2 Josh … Phil… East       13-J… G        6'6       200    26
 3 Jayle… Bost… East       30-D… SF       6'7       220    23
 4 Brand… New … West       30-D… SF       6'9       190    22
 5 Bam A… Miam… East       16-D… C        6'10      255    22
 6 Spenc… Broo… East       25-N… PG       6'6       210    26
 7 Luka … Dall… West       25-N… SF       6'7       218    21
 8 Pasca… Toro… East       11-N… F        6'9       230    25
 9 Trae … Atla… East       28-O… PG       6'2       180    21
10 Karl-… Minn… West       28-O… C        7'0       248    24
# … with 682 more rows, and 9 more variables: Draft_Year <dbl>,
#   Seasons_in_league <dbl>, Season <chr>, Season_short <dbl>,
#   Pre_draft_Team <chr>, Real_value <dbl>, Height_CM <dbl>,
#   Weight_KG <dbl>, Last_Season <dbl>

Arranging Data Frames by Variables

Another dplyr function, arrange() allows me to specify a variable I want to use to reorder rows of my data frame. Let’s say that I want to arrange my data so that the players’ draft year is in ascending order. I would type this:

# A tibble: 1,334 x 17
   Player Team  Conference Date  Position Height Weight   Age
   <chr>  <chr> <chr>      <chr> <chr>    <chr>   <dbl> <dbl>
 1 Rick … Hous… <NA>       10-F… F        6'7       205    35
 2 Elvin… Wash… <NA>       9-No… FC       6'9       235    35
 3 Karee… Los … <NA>       24-M… C        7'2       225    37
 4 Karee… Los … <NA>       8-Ap… C        7'2       225    36
 5 Karee… Los … <NA>       12-F… C        7'2       225    36
 6 Karee… Los … <NA>       28-F… C        7'2       225    34
 7 Karee… Los … <NA>       21-D… C        7'2       225    33
 8 Karee… Los … <NA>       17-F… C        7'2       225    32
 9 Karee… Los … <NA>       9-De… C        7'2       225    32
10 Tiny … Bost… <NA>       11-J… PG       6'1       150    32
# … with 1,324 more rows, and 9 more variables: Draft_Year <dbl>,
#   Seasons_in_league <dbl>, Season <chr>, Season_short <dbl>,
#   Pre_draft_Team <chr>, Real_value <dbl>, Height_CM <dbl>,
#   Weight_KG <dbl>, Last_Season <dbl>

Let’s say I change my mind and I want the players’ draft year to be in descending order. It would look like this:

# A tibble: 1,334 x 17
   Player Team  Conference Date  Position Height Weight   Age
   <chr>  <chr> <chr>      <chr> <chr>    <chr>   <dbl> <dbl>
 1 Luka … Dall… West       25-N… SF       6'7       218    21
 2 Trae … Atla… East       28-O… PG       6'2       180    21
 3 Trae … Atla… East       25-M… PG       6'2       180    20
 4 Bam A… Miam… East       16-D… C        6'10      255    22
 5 Donov… Utah… West       4-Ma… SG       6'3       215    22
 6 Donov… Utah… West       14-J… SG       6'3       215    22
 7 Ben S… Phil… East       20-J… PF       6'10      230    23
 8 Jayle… Bost… East       30-D… SF       6'7       220    23
 9 Brand… New … West       30-D… SF       6'9       190    22
10 Pasca… Toro… East       11-N… F        6'9       230    25
# … with 1,324 more rows, and 9 more variables: Draft_Year <dbl>,
#   Seasons_in_league <dbl>, Season <chr>, Season_short <dbl>,
#   Pre_draft_Team <chr>, Real_value <dbl>, Height_CM <dbl>,
#   Weight_KG <dbl>, Last_Season <dbl>

Whew, okay that was a lot! That’s it for data frames! Until next time…

Citation

For attribution, please cite this work as

Brantley (2020, Jan. 30). Data Sci Dani: Working With Data Frames. Retrieved from https://datascidani.com/posts/working_with_dataframes 01-30-20/

BibTeX citation

@misc{brantley2020working,
  author = {Brantley, Danielle},
  title = {Data Sci Dani: Working With Data Frames},
  url = {https://datascidani.com/posts/working_with_dataframes 01-30-20/},
  year = {2020}
}