Contents

Various Summarization Functions

Introduction

We have covered basic summarization operation here.

We will now explore some common summarization functions.

Procedure

We will be working with a custom dataframe.

 
# package for creating dataframe
library(tibble) 

# tibble or dataframe 
df <- tibble(numeric_col = as.integer(c(1:10)), 
             categorical_col = c("A","B","C","D","E","F","G","H","I","J")
             )
View(df)

Few rows of the data are:

summarization functions

  • Compute Mean

We will compute mean for the numerical column.

 
# refer procedure for definition of df
library(dplyr)

# compute mean for numeric_col
result <- summarize(df, mean=mean(numeric_col))
View(result)

The output of above code is:

mean

  • Compute Median

We will compute median for the numerical column.

 
# refer procedure for definition of df
library(dplyr)

# compute median for numeric_col
result <- summarize(df, median = median(numeric_col))
View(result)

The output of above code is:

median

  • Compute Standard Deviation

We will compute standard deviation for the numerical column. The standard deviation is computed using the squared distances formula.

 
# refer procedure for definition of df
library(dplyr)

# compute standard deviation for numeric_col
result <- summarize(df, `standard deviation` = sd(numeric_col))
View(result)

The output of above code is:

standard deviation

It means from the mean the values on average spread by 3.02765.

  • Compute Interquartile Range

We will compute interquartile range for the numerical column.

 
# refer procedure for definition of df
library(dplyr)

# compute interquartile range for numeric_col
result <- summarize(df, `interquartile range` = IQR(numeric_col))
View(result)

The output of above code is:

IQR

  • Compute Median absolute deviation

We will compute median absolute deviation for the numerical column.

 
# refer procedure section for definition of df
library(dplyr)

# compute median absolute deviation for numeric_col
result <- summarize(df, `median absolute deviation` = mad(numeric_col))
View(result)

The output of above code is:

mad

  • Find Minimum

We will find minimum value for the numeric column as well as the categorical column. For categorical column the minimum is found by using the Lexicographical Order.

 
# refer procedure section for definition of df
library(dplyr)

# find min for numeric_col and categorical_col
result <- summarize(df, `minimum_numeric` = min(numeric_col),
                    `minimum_categorical` = min(categorical_col))
View(result)

The output of above code is:

min

  • Find Quantile

We will compute the 25th quantile and the 50th quantile for numerical column.

 
# refer procedure section for definition of df
library(dplyr)

# find 25th and 50th quantile for numeric_col
result <- summarize(df, `25th Quantile` = quantile(numeric_col,0.25),`50th Quantile or Median` = quantile(numeric_col,0.5))
View(result)

The output of above code is:

quantile

  • Find Maximum

We will find maximum value for the numeric column as well as the categorical column. For categorical column the maximum is found by using the Lexicographical Order.

 
# refer procedure section for definition of df
library(dplyr)

# find max for numeric_col and categorical_col
result <- summarize(df, `maximum_numeric` = max(numeric_col),
                    `maximum_categorical` = max(categorical_col))
View(result)

The output of above code is:

max

  • Find First

We will find first value for the numeric column as well as the categorical column. For categorical column the first value is found by using the Lexicographical Order.

 
# refer procedure section for definition of df
library(dplyr)

# find first for numeric_col and categorical_col
result <- summarize(df, 
                    `first_numeric` = first(numeric_col),
                    `first_categorical` = first(categorical_col))
View(result)

The output of above code is:

first

  • Find nth

We will find nth value for the numeric column as well as the categorical column. For categorical column the nth value is found by using the Lexicographical Order.

 
# refer procedure section for definition of df
library(dplyr)

# find 3rd value for numeric_col and categorical_col
result <- summarize(df, 
                    `3rd_numeric` = nth(numeric_col,3),
                    `3rd_categorical` = nth(categorical_col,3))
View(result)

The output of above code is:

nth

  • Find Last

We will find last value for the numeric column as well as the categorical column. For categorical column the last value is found by using the Lexicographical Order.

 
# refer procedure section for definition of df
library(dplyr)

# find last for numeric_col and categorical_col
result <- summarize(df, 
                    `last_numeric` = last(numeric_col),
                    `last_categorical` = last(categorical_col))
View(result)

The output of above code is:

last

  • Count

We will count the number of rows in df dataframe.

 
# refer procedure section for definition of df
library(dplyr)

# count rows in df
result <- summarize(df, `row count` = n())
View(result)

The output of above code is:

count

  • Count unique

We will count the number of unique or distinct values in numeric_col and categorical_col.

 
# refer procedure section for definition of df
library(dplyr)

# count unique values in numeric_col and categorical_col
result <- summarize(df, 
                    `numeric_col unique` = n_distinct(numeric_col),
                    `categorical_col unique` = n_distinct(categorical_col)
                    )
View(result)

The output of above code is:

count unique

  • Count non-missing

We will count the number of non-missing values in numeric_col and categorical_col. Non-missing values are NA’s.

 
# refer procedure section for definition of df
library(dplyr)

# count non-missing values in numeric_col and categorical_col
result <- summarize(df, 
                    `numeric_col non-missing values` = sum(!is.na(numeric_col)),
                    `categorical_col non-missing values` = sum(!is.na(categorical_col))
                    )
View(result)

The output of above code is:

count non-missing

Conclusion

Thus we have successfully explored various summarization functions.

References

  • https://r4ds.had.co.nz/