Two Simple Ways to Fill Dummy Data into the Right Rows

Two Simple Ways to Fill Dummy Data into the Right Rows.

---

Introduction

When preparing statistical summaries, we often need to create dummy data to maintain consistent structures.For example, ensuring all combinations of categorical variables appear in a summary table. This post introduces two simple approaches to fill dummy rows into your dataset using R and the tidyverse.

---

Example Dataset

Let’s start by creating a sample dataset:

library(tidyverse)

── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.4     ✔ readr     2.1.5
✔ forcats   1.0.1     ✔ stringr   1.5.2
✔ ggplot2   4.0.0     ✔ tibble    3.3.0
✔ lubridate 1.9.4     ✔ tidyr     1.3.1
✔ purrr     1.1.0     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

set.seed(2025)

temp1 <- tibble(
  treatment = sample(c("g1", "g2"), size = 10, replace = TRUE),
  sub       = sample(c("x1", "x2", "x3"), size = 10, replace = TRUE),
  id        = 1:10
)

temp1

# A tibble: 10 × 3
   treatment sub      id
   <chr>     <chr> <int>
 1 g1        x1        1
 2 g2        x3        2
 3 g2        x3        3
 4 g2        x3        4
 5 g1        x1        5
 6 g1        x3        6
 7 g2        x3        7
 8 g1        x3        8
 9 g2        x3        9
10 g1        x2       10

This dataset contains three variables:

• treatment with values “g1” and “g2”,

• sub with values “x1”, “x2”, and “x3”,

• id as a sequential index.

👉 Now imagine we need to count records by treatment and sub, but we also want to include a new category — y1 — under sub, even if it doesn’t appear in the data. How can we achieve that?

Method 1 — Using crossing()

This approach manually builds a dummy dataset and merges it with the existing summary.

Steps:

1️⃣ Obtain unique treatment values.

2️⃣ Create all desired combinations using crossing().

3️⃣ Combine the dummy and original counts.

# Count the original combinations

temp2 <- temp1 %>%
count(treatment, sub, name = "count")

# Extract unique treatments

temp2_nodup <- temp2 %>%
  arrange(treatment, sub, count) %>%
  group_by(treatment) %>%
  mutate(id = row_number()) %>%
  filter(row_number() == 1) %>%
  select(treatment) %>%
  ungroup()


# Create dummy data with all sub categories, including y1

dummy <- crossing(
  temp2_nodup,
  sub = c("x1", "x2", "x3", "y1"),
  count = 0
)

# Combine and keep the latest record per combination

temp3 <- temp2 %>%
  rbind(dummy) %>%
  arrange(treatment, sub, count) %>%
  group_by(treatment, sub) %>%
  mutate(id1 = row_number(), id2 = n()) %>%
  filter(row_number() == n()) %>%
  select(-matches("^id")) %>%
  ungroup()

temp3

# A tibble: 8 × 3
  treatment sub   count
  <chr>     <chr> <dbl>
1 g1        x1        2
2 g1        x2        1
3 g1        x3        2
4 g1        y1        0
5 g2        x1        0
6 g2        x2        0
7 g2        x3        5
8 g2        y1        0

Method 2 — Using complete()

A cleaner and more concise way uses tidyr::complete(), which automatically fills missing combinations.

Steps:

1️⃣ Count existing records by group.

2️⃣ Use complete() to add the missing combinations.

---

way2 <- temp1 %>%
  count(treatment, sub, name = "count") %>%
  complete(
    treatment,
    sub = c("x1", "x2", "x3", "y1"),
    fill = list(count = 0)
    )

way2 

# A tibble: 8 × 3
  treatment sub   count
  <chr>     <chr> <int>
1 g1        x1        2
2 g1        x2        1
3 g1        x3        2
4 g1        y1        0
5 g2        x1        0
6 g2        x2        0
7 g2        x3        5
8 g2        y1        0

🎨Conclusion

This is a common requirement in data processing — ensuring all expected categories appear in your summaries or reports. Both crossing() and complete() from the tidyverse provide efficient ways to generate dummy data and maintain structural integrity.

• Use crossing() when you need full control over combinations and want to manually build the structure.

• Use complete() for a more concise, declarative approach that integrates naturally into tidy pipelines.

In short, crossing() gives you full control to build combinations manually, while complete() offers a cleaner, automatic way to fill missing categories in your data.

---