Tidy Tuesday: Repair Cafes Worldwide

tidytuesday
R
sustainability
repair
Exploring repair success rates and failure modes across product categories at repair cafes worldwide.
Author

Sean Thimons

Published

April 7, 2026

Preface

From TidyTuesday repository.

Repair Cafes are free meeting places where visitors can bring their broken items to be repaired by skilled volunteers. This dataset comes from the Repair Cafe Foundation and covers repairs logged from hundreds of cafes worldwide, including the item category, repair outcome, and failure reasons when a repair could not be completed.

Loading necessary packages

My handy booster pack that allows me to install (if needed) and load my usual and favorite packages, as well as some helpful functions.

Code 
# Packages ----------------------------------------------------------------

{
  # Install pak if it's not already installed
  if (!requireNamespace("pak", quietly = TRUE)) {
    install.packages("pak", repos = "https://cloud.r-project.org")
  }

  # CRAN Packages ----
  install_booster_pack <- function(package, load = TRUE) {
    for (pkg in package) {
      if (!requireNamespace(pkg, quietly = TRUE)) {
        pak::pkg_install(pkg)
      }
      if (load) {
        library(pkg, character.only = TRUE)
      }
    }
  }

  core_packages <- c(
    'fs',
    'here',
    'janitor',
    'rio',
    'tidyverse',
    'skimr',
    'paletteer',
    'tidytuesdayR'
  )

  optional_packages <- c('ggtext', 'glue')

  install_booster_pack(package = unique(c(core_packages, optional_packages)), load = TRUE)
  rm(install_booster_pack, core_packages, optional_packages)

  # Custom Functions ----

  `%ni%` <- Negate(`%in%`)

  geometric_mean <- function(x) {
    exp(mean(log(x[x > 0]), na.rm = TRUE))
  }

  my_skim <- skim_with(
    numeric = sfl(
      n = length,
      min = ~ min(.x, na.rm = T),
      p25 = ~ stats::quantile(., probs = .25, na.rm = TRUE, names = FALSE),
      med = ~ median(.x, na.rm = T),
      p75 = ~ stats::quantile(., probs = .75, na.rm = TRUE, names = FALSE),
      max = ~ max(.x, na.rm = T),
      mean = ~ mean(.x, na.rm = T),
      geo_mean = ~ geometric_mean(.x),
      sd = ~ stats::sd(., na.rm = TRUE),
      hist = ~ inline_hist(., 5)
    ),
    append = FALSE
  )
}

Load raw data from package

raw <- tidytuesdayR::tt_load('2026-04-07')

repairs      <- raw$repairs
repairs_text <- raw$repairs_text

Exploratory Data Analysis

The my_skim() function is a modified version of the skimr::skim() function that returns the number of missing data points (cells as NA) as well as the inverse (e.g.: number of rows that are not NA), the count, minimum, 25%, median, 75%, max, mean, geometric mean, and standard deviation. It also generates a little ASCII histogram. Neat!

repairs

my_skim(repairs)
Data summary
Name repairs
Number of rows 178749
Number of columns 11
_______________________
Column type frequency:
character 7
Date 1
numeric 3
________________________
Group variables None

Variable type: character

skim_variable n_missing complete_rate min max empty n_unique whitespace
repair_id 0 1 16 25 0 178716 0
repair_cafe_name 5 1 3 46 0 447 0
country 0 1 2 2 0 25 0
kind_of_product 35 1 3 52 0 1951 0
category 5 1 5 33 0 14 0
brand 147 1 2 38 0 8600 0
repaired 2 1 2 4 0 4 0

Variable type: Date

skim_variable n_missing complete_rate min max median n_unique
repair_date 0 1 2015-04-04 2026-08-07 2024-01-31 2582

Variable type: numeric

skim_variable n_missing complete_rate n min p25 med p75 max mean geo_mean sd hist
repair_cafe_number 5 1.00 178749 2 50 300 541 997 321.11 150.34 274.97 ▇▃▃▃▁
estimated_year_of_production 114383 0.36 178749 -6 2002 2014 2019 2100 2000.21 2007.29 124.58 ▁▁▁▁▇
repairability 53885 0.70 178749 0 5 7 9 10 6.66 5.79 2.74 ▃▃▅▇▇ 
# Check exact values in key categorical columns before any filtering
cat("\n--- repaired column values ---\n")

--- repaired column values ---
repairs |> count(repaired, sort = TRUE) |> print()
# A tibble: 5 × 2
  repaired      n
  <chr>     <int>
1 yes      112775
2 no        43112
3 half      22859
4 <NA>          2
5 ja            1
cat("\n--- category column values ---\n")

--- category column values ---
repairs |> count(category, sort = TRUE) |> print()
# A tibble: 15 × 2
   category                              n
   <chr>                             <int>
 1 Household appliances electric     70983
 2 Textile                           24412
 3 Display and sound equipment       20415
 4 Other                             10435
 5 Tools electric                    10340
 6 Computer equipment / phones        9488
 7 Clocks / alarm clocks              7016
 8 Bicycles                           6252
 9 Household appliances non-electric  5593
10 Tools non-electric                 4178
11 Jewelry                            2852
12 Toys electric                      2645
13 Furniture                          2120
14 Toys non-electric                  2015
15 <NA>                                  5
cat("\n--- country column values ---\n")

--- country column values ---
repairs |> count(country, sort = TRUE) |> print(n = 30)
# A tibble: 25 × 2
   country     n
   <chr>   <int>
 1 NL      90839
 2 GB      37692
 3 FR      19724
 4 CA       7642
 5 AU       7079
 6 DE       4509
 7 US       3585
 8 BE       1975
 9 ES       1711
10 IL       1559
11 AT       1248
12 NZ        437
13 IE        242
14 IT        139
15 FI        107
16 TW         78
17 LI         56
18 PT         31
19 GF         27
20 LU         26
21 KR         23
22 ZA         13
23 HK          4
24 CH          2
25 SE          1

repairs_text

my_skim(repairs_text)
Data summary
Name repairs_text
Number of rows 178749
Number of columns 12
_______________________
Column type frequency:
character 12
________________________
Group variables None

Variable type: character

skim_variable n_missing complete_rate min max empty n_unique whitespace
repair_id 0 1.00 16 25 0 178716 0
model 115064 0.36 1 132 0 37802 0
defect_found 30903 0.83 1 255 0 80236 0
problem_description 52079 0.71 1 900 0 80694 0
repair_method 94499 0.47 1 5943 0 57198 0
partial_repair_notes 159497 0.11 1 1304 0 17715 0
failure_reasons 152047 0.15 12 43 0 10 0
failure_reason_open 172417 0.04 1 328 0 5051 0
used_repair_info 69912 0.61 2 7 0 3 0
repair_info_source 175339 0.02 1 629 0 1627 0
repair_info_url 173200 0.03 4 23 0 25 0
suggestions 173169 0.03 1 790 0 4832 0 
cat("\n--- failure_reasons column values ---\n")

--- failure_reasons column values ---
repairs_text |> count(failure_reasons, sort = TRUE) |> print()
# A tibble: 11 × 2
   failure_reasons                                  n
   <chr>                                        <int>
 1 <NA>                                        152047
 2 No way to fix the product                     5563
 3 Spare parts not available at repair session   4750
 4 Unidentified failure                          3676
 5 Too worn out                                  3352
 6 No way to open the product                    2311
 7 Spare parts not available on the market       2190
 8 Spare parts too expensive                     2179
 9 Too time consuming                            1233
10 Lack of equipment at repair location           878
11 Repair information not available               570

The repairs table has ~179,000 repair records spanning hundreds of repair cafes worldwide. The dominant countries are the Netherlands (NL, ~91k records), Great Britain (GB, ~38k), and France (FR, ~20k), reflecting where the Repair Cafe movement originated. The repaired column uses “yes”, “no”, and “half” as its primary outcome values, with a small number of “ja” entries (Dutch for “yes”). The repairability score runs 0–10 and is missing for roughly a third of records.

Repair Success by Product Category

The core question for any repair cafe visitor is: “What are my chances of walking out with a fixed item?” I look at repair outcomes across all product categories, normalizing for the mix of “yes”, “half”, and “no” results. The “half” outcome — a partial fix — is treated as a mid-point rather than a failure.

# Normalize repaired values: treat "ja" (Dutch yes) as "yes"
repairs_clean <- repairs |>
  mutate(
    repaired_norm = case_when(
      repaired %in% c("yes", "ja") ~ "Repaired",
      repaired == "half"           ~ "Partial",
      repaired == "no"             ~ "Not repaired",
      TRUE                         ~ NA_character_
    )
  ) |>
  filter(!is.na(category), !is.na(repaired_norm))

# Verify we have data
stopifnot(nrow(repairs_clean) > 0)
cat("Rows after cleaning:", nrow(repairs_clean), "\n")
Rows after cleaning: 178742 
# Count outcomes per category
category_outcomes <- repairs_clean |>
  count(category, repaired_norm) |>
  group_by(category) |>
  mutate(
    total      = sum(n),
    pct        = n / total
  ) |>
  ungroup()

# Verify
stopifnot(nrow(category_outcomes) > 0)
cat("Categories x outcomes:\n")
Categories x outcomes:
print(category_outcomes)
# A tibble: 42 × 5
   category                    repaired_norm     n total    pct
   <chr>                       <chr>         <int> <int>  <dbl>
 1 Bicycles                    Not repaired    571  6252 0.0913
 2 Bicycles                    Partial         576  6252 0.0921
 3 Bicycles                    Repaired       5105  6252 0.817 
 4 Clocks / alarm clocks       Not repaired   1656  7016 0.236 
 5 Clocks / alarm clocks       Partial        1035  7016 0.148 
 6 Clocks / alarm clocks       Repaired       4325  7016 0.616 
 7 Computer equipment / phones Not repaired   2802  9488 0.295 
 8 Computer equipment / phones Partial        2285  9488 0.241 
 9 Computer equipment / phones Repaired       4401  9488 0.464 
10 Display and sound equipment Not repaired   7368 20414 0.361 
# ℹ 32 more rows
# Order categories by % fully repaired
cat_order <- category_outcomes |>
  filter(repaired_norm == "Repaired") |>
  arrange(pct) |>
  pull(category)

category_outcomes <- category_outcomes |>
  mutate(category = factor(category, levels = cat_order))
pal_colors <- as.character(paletteer::paletteer_d("lisa::OskarSchlemmer", n = 5))
# Assign outcome colors: Repaired = warm gold, Partial = mid blue-purple, Not repaired = dark red
outcome_colors <- c(
  "Repaired"     = pal_colors[3],   # DABD61 warm gold
  "Partial"      = pal_colors[2],   # 8785B2 muted purple
  "Not repaired" = pal_colors[5]    # BE3428 warm red
)

p <- ggplot(
  category_outcomes,
  aes(x = pct, y = category, fill = repaired_norm)
) +
  geom_col(width = 0.7, position = "stack") +
  scale_x_continuous(
    labels = scales::percent_format(accuracy = 1),
    expand = expansion(mult = c(0, 0.02))
  ) +
  scale_fill_manual(
    values = outcome_colors,
    breaks = c("Repaired", "Partial", "Not repaired"),
    name   = "Outcome"
  ) +
  geom_vline(xintercept = 0.5, linetype = "dashed", color = "grey40", linewidth = 0.5) +
  labs(
    title    = "Repair success rates at repair cafes worldwide",
    subtitle = "Share of items fully repaired, partially fixed, or not repaired — by product category",
    x        = "Share of repairs",
    y        = NULL,
    caption  = "Source: Repair Cafe Foundation via TidyTuesday 2026-04-07 | 'half' outcomes treated as partial fixes"
  ) +
  theme_minimal(base_size = 13) +
  theme(
    plot.title       = element_text(face = "bold", size = 15),
    plot.subtitle    = element_text(color = "grey30", size = 11, margin = margin(b = 10)),
    plot.caption     = element_text(color = "grey50", size = 9, hjust = 0),
    legend.position  = "top",
    legend.title     = element_text(face = "bold"),
    panel.grid.major.y = element_blank(),
    panel.grid.minor   = element_blank(),
    axis.text.y      = element_text(size = 11)
  )

p

# Join repairs with repairs_text to get failure reasons
repairs_with_text <- repairs |>
  inner_join(repairs_text |> select(repair_id, failure_reasons), by = "repair_id") |>
  filter(!is.na(failure_reasons), !is.na(category))

stopifnot(nrow(repairs_with_text) > 0)
cat("Rows with failure reasons:", nrow(repairs_with_text), "\n")
Rows with failure reasons: 26720 
# Top failure reasons overall
failure_summary <- repairs_with_text |>
  count(failure_reasons, sort = TRUE) |>
  filter(failure_reasons != "NA") |>
  slice_max(n, n = 9) |>
  mutate(
    failure_reasons = str_wrap(failure_reasons, width = 35),
    failure_reasons = fct_reorder(failure_reasons, n)
  )

stopifnot(nrow(failure_summary) > 0)

pal_fail <- as.character(paletteer::paletteer_d("lisa::OskarSchlemmer", n = 5))

p2 <- ggplot(failure_summary, aes(x = n, y = failure_reasons, fill = n)) +
  geom_col(width = 0.7, show.legend = FALSE) +
  scale_fill_gradient(low = pal_fail[2], high = pal_fail[4]) +
  scale_x_continuous(
    labels = scales::comma,
    expand = expansion(mult = c(0, 0.05))
  ) +
  labs(
    title    = "Why do repairs fail at repair cafes?",
    subtitle = "Most common logged reasons when a repair could not be completed",
    x        = "Number of repair records",
    y        = NULL,
    caption  = "Source: Repair Cafe Foundation via TidyTuesday 2026-04-07"
  ) +
  theme_minimal(base_size = 13) +
  theme(
    plot.title       = element_text(face = "bold", size = 15),
    plot.subtitle    = element_text(color = "grey30", size = 11, margin = margin(b = 10)),
    plot.caption     = element_text(color = "grey50", size = 9, hjust = 0),
    panel.grid.major.y = element_blank(),
    panel.grid.minor   = element_blank(),
    axis.text.y      = element_text(size = 11)
  )

p2

Final thoughts and takeaways

The repair cafe dataset paints an encouraging picture of the repair movement. Across roughly 179,000 logged repair attempts:

  • Textiles and non-electric tools tend to have the highest full-repair rates. These categories often involve straightforward fixes — sewing a seam, sharpening a blade — that skilled volunteers can reliably complete within a session.

  • Bicycles and clocks/alarm clocks also perform well. The mechanical nature of these items, combined with the specialized volunteers often present at repair cafes, makes successful repair relatively common.

  • Electronic and computer equipment sits in the middle. While volunteers can fix many issues, component-level repairs require parts and equipment that aren’t always on hand.

  • The dominant failure modes are structural rather than skill-related: “No way to fix the product” (e.g., items that are sealed or glued shut) and “Spare parts not available” together account for the largest share of unsuccessful repairs. This points directly at a design and supply chain problem — products engineered without repairability in mind.

The repair cafe model is fundamentally optimistic: bring what’s broken, and skilled community members will try their best. But the data reveals where that optimism runs into systemic barriers. The right-to-repair movement’s emphasis on access to spare parts and service documentation maps directly onto the failure reasons most common in this dataset.