2.3.6. Summarizing Data

1. Introduction

Summarizing data is a key part of data wrangling and analysis. Whether for reporting, checking data quality, or exploring your dataset, R provides powerful tools to quickly summarize and understand your data. This chapter covers essential summary functions from dplyr, janitor, and skimr.

2. Why Summarize Data?

• Quickly understand the structure and content of your dataset.
• Check for missing values, duplicates, or outliers.
• Produce summary statistics for reports or presentations.
• Aggregate data for modeling or visualization.
• Validate results after data wrangling steps.

3. Summarizing with summarize()

• The summarize() (or summarise()) function from dplyr creates summary statistics for your data.
• Use with or without group_by() for overall or grouped summaries.

Input Table:

R Code:

dm %>%
  summarize(N = n())

• summarize(N = n()) counts the number of rows in the dataset and returns it as column N.

Output Table:

• Explanation:
  • Counts the number of rows in the dataset.

4. Grouped Summaries with group_by() and summarize()

• Use group_by() to summarize within categories.

R Code:

dm %>%
  group_by(ARM) %>%
  summarize(N = n())

• group_by(ARM) groups the data by the ARM column.
• summarize(N = n()) counts the number of rows in each group.

Output Table:

• Explanation:
  • Counts the number of subjects in each treatment arm.

5. Summarizing with Other Functions

• You can use functions like mean(), median(), min(), max(), and sum() inside summarize().

R Code:

dm %>%
  group_by(ARM) %>%
  summarize(
    N = n(),
    mean_age = mean(AGE),
    min_age = min(AGE),
    max_age = max(AGE)
  )

• mean_age = mean(AGE) calculates the mean age for each group.
• min_age = min(AGE) and max_age = max(AGE) find the minimum and maximum age in each group.

Output Table:

• Explanation:
  • Provides count, mean, minimum, and maximum age for each treatment arm.

6. Quick Counts with tally() and count()

• tally() counts rows or sums a numeric column.
• count() counts unique values of one or more columns.

R Code:

dm %>% tally()

• tally() returns the total number of rows in the dataset.

Output Table:

R Code:

dm %>% count(SEX)

• count(SEX) counts the number of occurrences for each unique value in the SEX column.

Output Table:

R Code:

dm %>% count(SEX, ARM)

• count(SEX, ARM) counts the number of occurrences for each combination of SEX and ARM.

Output Table:

7. Adding Counts to Data with add_tally() and add_count()

• add_tally() adds a column with the total row count to each row.
• add_count() adds a column with the group count to each row.

R Code:

dm %>% add_tally()

• add_tally() adds a column n to every row, showing the total number of rows in the dataset.

Output Table:

R Code:

dm %>% add_count(ARM)

• add_count(ARM) adds a column n to every row, showing the count for that row's ARM group.

Output Table:

8. Summarizing Categorical Data with tabyl()

• The tabyl() function from janitor quickly summarizes categorical variables.

R Code:

library(janitor)
dm %>% tabyl(ARM)

• tabyl(ARM) returns a frequency table for the ARM column, including counts and proportions.

Output Table:

• Explanation:
  • Shows counts and proportions for each category.

9. Summarizing Numeric Data with summary() and skim()

• summary() (base R) gives min, max, mean, quartiles for numeric columns.
• skim() from skimr provides a detailed summary for all columns.

R Code:

summary(dm$AGE)

• summary(dm$AGE) returns min, 1st quartile, median, mean, 3rd quartile, and max for AGE.

Output Table:

R Code:

library(skimr)
skim(dm)

• skim(dm) provides a comprehensive summary for each variable, including missingness, mean, sd, and quantiles.

Output Table:

── Data Summary ────────────────────────
                           Values
Name                       dm    
Number of rows             4     
Number of columns          4     
_______________________          
Column type frequency:           
  character                3     
  numeric                  1     
________________________         
Group variables            None  

── Variable type: character ──────────────────────────────────────────────────────────────────────
  skim_variable n_missing complete_rate min max empty n_unique whitespace
1 USUBJID               0             1  10  10     0        4          0
2 SEX                   0             1   1   1     0        2          0
3 ARM                   0             1   6   7     0        2          0

── Variable type: numeric ────────────────────────────────────────────────────────────────────────
  skim_variable n_missing complete_rate mean   sd p0  p25 p50  p75 p100 hist 
1 AGE                   0             1   46 17.7 29 32.8  44 57.2   67 ▇▁▁▃▃

• Explanation:
  • skim() provides a comprehensive summary for each variable, including missingness and distribution.

10. Finding Duplicates with get_dupes()

• The get_dupes() function from janitor helps identify duplicate records.

Input Table:

R Code:

library(janitor)
dm %>% get_dupes(USUBJID)

• get_dupes(USUBJID) finds rows where USUBJID is duplicated and adds a dupe_count column.

Output Table:

11. Exploring More: Advanced Summaries

Below are advanced ways to summarize your data, with detailed explanations, input, and output examples.

• Multiple summaries at once with across()

  You can use across() inside summarize() to apply several summary functions to one or more columns at once.

  Input Table:

  USUBJID	AGE
  01-701-101	34
  01-701-102	29
  01-701-103	67
  01-701-104	54

  R Code:

  dm %>%
    summarize(across(AGE, list(mean = mean, sd = sd, min = min, max = max)))
  

  • across(AGE, list(mean = mean, sd = sd, min = min, max = max)) applies each function to the AGE column.
  • The result is a single-row summary with columns for mean, sd, min, and max.

  Output Table:

  AGE_mean	AGE_sd	AGE_min	AGE_max
  46	16.6	29	67

• Summarize after grouping by multiple variables

  You can group by more than one variable to get summaries for each combination.

  Input Table:

  USUBJID	AGE	SEX	ARM
  01-701-101	34	M	Placebo
  01-701-102	29	F	Drug X
  01-701-103	67	F	Placebo
  01-701-104	54	M	Drug X

  R Code:

  dm %>%
    group_by(SEX, ARM) %>%
    summarize(N = n(), mean_age = mean(AGE))
  

  • group_by(SEX, ARM) groups the data by both SEX and ARM.
  • summarize(N = n(), mean_age = mean(AGE)) gives the count and mean age for each group.

  Output Table:

  SEX	ARM	N	mean_age
  F	Drug X	1	29
  F	Placebo	1	67
  M	Drug X	1	54
  M	Placebo	1	34

• Summarize with custom functions

  You can use any function inside summarize(), such as median().

  Input Table:

  USUBJID	AGE
  01-701-101	34
  01-701-102	29
  01-701-103	67
  01-701-104	54

  R Code:

  dm %>%
    summarize(median_age = median(AGE, na.rm = TRUE))
  

  • Calculates the median age, ignoring missing values.

  Output Table:

  median_age
  44

• Summarize with weighted means

  Sometimes you want to calculate a mean that gives different weights to different rows.

  Input Table:

  USUBJID	AGE	WEIGHT
  01-701-101	34	1
  01-701-102	29	2
  01-701-103	67	1
  01-701-104	54	1

  R Code:

  dm %>%
    summarize(weighted_mean = weighted.mean(AGE, w = WEIGHT))
  

  • weighted.mean(AGE, w = WEIGHT) computes the mean of AGE, weighting each row by the WEIGHT column.

  Output Table:

  weighted_mean
  42.6

• Summarize with proportions

  You can calculate the proportion of each group using count() and mutate().

  Input Table:

  USUBJID	ARM
  01-701-101	Placebo
  01-701-102	Drug X
  01-701-103	Placebo
  01-701-104	Drug X

  R Code:

  dm %>%
    count(ARM) %>%
    mutate(prop = n / sum(n))
  

  • count(ARM) counts the number of rows for each ARM.
  • mutate(prop = n / sum(n)) calculates the proportion for each ARM.

  Output Table:

  ARM	n	prop
  Drug X	2	0.5
  Placebo	2	0.5

• Explanation for all above:
  • These advanced summaries allow you to quickly get deeper insights into your data.
  • You can combine grouping, multiple summary functions, custom logic, and proportions for flexible reporting and analysis.
  • Input and output tables help you see exactly what each code block does.

12. Conclusion

• Summarizing data is essential for understanding, validating, and reporting your data.
• Use summarize(), count(), tabyl(), skim(), and related functions for flexible, powerful summaries.
• Combine with group_by() for grouped summaries.
• Explore advanced summaries for deeper insights and better data quality.

**Resource download links**

2.3.6.-Summarizing-Data.zip