How many “Big” Saturday Anniversaries Does Anyone Get?

The first question to explore is for the 834 Saturdays in our data as potential wedding dates, how many of the “big” anniversaries will fall on a Saturday. The following code block is pretty vanilla dplyr with the use of count() and mutate(). If you’ve never seen the glue() package and function before, it works a lot like paste() in its most basic form. The main difference is that R will execute the code within the { } so it can be included directly within the quotes rather than separated by commas. It can also be used similar to .format() in Python.

wedding_dates_w_annv %>% 
  # Get frequencies of Big Saturday Anniversaries
  count(num_big_sat) %>% 
  # Create %s 
  mutate(pct = n/sum(n)) %>% 
  ggplot(aes(x = as.factor(num_big_sat), y = pct, fill = as.factor(num_big_sat))) +
    geom_col() + 
    geom_text(aes(label = glue("{pct %>% scales::percent()} (n={n %>% scales::comma()})")), nudge_y = 0.02) + 
    labs(title = "How many ***BIG*** anniversaries are celebrated on Saturday?",
         subtitle = glue("Saturday Wedding Dates 2010 - 2025 (n = {nrow(wedding_dates_w_annv)})"),
         caption = "Big = Multiple of 5 (5th, 10th, etc.)",
         x = "# of Big Anniversaries on Saturdays",
         y = "% of Wedding Dates") + 
    scale_fill_discrete(guide = F) + 
    cowplot::theme_cowplot() + 
    theme(
      plot.title = ggtext::element_markdown(),
      axis.text.y = element_blank(),
      axis.ticks.y = element_blank()
    )

The primary reason there isn’t a best or worst wedding date is that all potential wedding dates either have 1 or 2 BIG anniversaries on a Saturday. So there isn’t too much of a difference in choice of dates.

So let’s look at how many anniversaries in total occur on a Saturday.

How many total Saturday Anniversaries Does Anyone Get?

wedding_dates_w_annv %>% 
  count(num_sat) %>% 
  mutate(pct = n/sum(n)) %>% 
  ggplot(aes(x = as.factor(num_sat), y = pct, fill = as.factor(num_sat))) +
  geom_col() + 
  geom_text(aes(label = glue("{pct %>% scales::percent()} (n={n %>% scales::comma()})")), nudge_y = 0.02) + 
  labs(title = "How many anniversaries are celebrated on Saturday?",
       subtitle = glue("Saturday Wedding Dates 2010 - 2025 (n = {nrow(wedding_dates_w_annv)})"),
       x = "# of Anniversaries on Saturdays",
       y = "% of Wedding Dates") + 
  scale_fill_discrete(guide = F) + 
  cowplot::theme_cowplot() + 
  theme(
    axis.text.y = element_blank(),
    axis.ticks.y = element_blank()
  )

Continuing with the theme of there not being major differences. 75% of Saturday Wedding dates will have 7 anniversaries on a Saturday and 25% will have 6. So, while 7 would be preferable, the difference between 7 vs. 6 again is not large.

Looking at Both Total Saturdays and “Big” Saturdays

Since “Big” Anniversaries had a 50/50 distribution and overall Saturdays had a 25/75 distribution the next step would be to see the cross-product of the two previous fields:

wedding_dates_w_annv %>% 
  count(num_sat, num_big_sat) %>% 
  mutate(pct = n / sum(n)) %>% 
  ggplot(aes(x = factor(num_sat), y = factor(num_big_sat), fill = pct)) + 
    geom_tile() + 
    geom_text(aes(label = glue("{pct %>% scales::percent()} \n (n={n})"))) + 
    labs(title = "A deeper look into Saturday Anniversaries",
         subtitle = glue("Saturday Wedding Dates 2010 - 2025 (n = {nrow(wedding_dates_w_annv)})"),
         x = "# of Anniversaries on Saturdays",
         y = "# of 'Big' Anniversaries on Saturday") + 
    scale_fill_gradient(guide = F, low = "#769293", high = "#fad7d5") +
    cowplot::theme_cowplot()

Looking across both dimensions, everyone who has two “big” anniversaries on a Saturday ALSO has 7 anniversaries on a Saturday. However, not everyone who has 7 anniversaries on Saturday will have 2 “big” anniversaries on a Saturday. Instead there are three groups:

• 6 Total / 1 Big (25%)
• 7 Total / 1 Big (25%)
• 7 Total / 2 Big (50%)

In this case, having 7 anniversaries and 2 “Big” Anniversaries seems preferable to the other two groups… if you only cared about having your anniversary on a Saturday.

What “Big” Anniversaries Will Be Celebrated on Saturdays?

So far, I’ve defined “big” anniversaries as multiples of 5 (5th, 10th, … 45th, 50th). However, I haven’t looked at which of those big ones are occurring on a Saturday. To show these “big” anniversaries I’ll use the eulerr package to create a Venn-Diagram of these years.

The package expects a specific format where each column is a logical indicating whether or not an observation is a member of that group. From a quick check on the big_sat_years field I can see that the only “big” anniversaries that fall on Saturdays are the 5th, 45th, and 50th.

Of note is the regular expression “\b5\b” for identifying the 5th anniversary. \\b represents a word boundary so it is included to make sure that the 5th anniversary doesn’t accidentally get picked up by str_detect() as part of 45 or 50, which would occur if I only searched for “5”.

wedding_dates_w_annv %>% 
  #Constructing Logicals for Venn Diagrams
  transmute(
    `5th \n Anniversary` = str_detect(big_sat_years, '\\b5\\b'),
    `45th \n Anniversary` = str_detect(big_sat_years, '45'),
    `50th \n Anniversary` = str_detect(big_sat_years, '50')
  ) %>%
  #Plot the Venn-Diagram
  euler() %>% 
  plot(quantities = list(type = c('counts', 'percent')),
       percentages = TRUE,
       main = "Which 'big' anniversaries get celebrated on Saturdays?",
       )

So 75% of wedding dates will celebrate their 45th anniversary on a Saturday. 50% will celebrate ONLY their 45th anniversary and 25% will celebrate their 45th and 50th anniversaries on a Saturday. The last 25% will celebrate their 5th and 50th anniversary on a Saturday. No one will ONLY celebrate either their 5th or 50th. Fitting this into our three group paradigm from the prior section:

• 6 Total / 1 Big (25%) – Will ONLY celebrate their 45th Anniversary
• 7 Total / 1 Big (25%) – Will ONLY celebrate their 45th Anniversary
• 7 Total / 2 Big (50%)
  • 25% will celebrate their 5th and 50th
  • 25% will celebrate their 45th and 50th

Is there a time component to which group you end up in?

This final section looks at the time component to whether you wind up in the 6/1, 7/1, or 7/2 group. In order to summarize to a Year/Month level, the average number of Saturdays and “Big” Saturdays will be used. Then in the following heat-map, the year of the wedding date appears on the y-axis and the month of the wedding is on the x-axis.

wedding_dates_w_annv %>% 
  #Reformat to Year-Month (%Y = Year w/ Century, %m = Month as Zero-Padded Decimal)
  mutate(
    m = month(wedding_date),
    y = year(wedding_date),
  ) %>% 
  group_by(m, y) %>% 
  #Get Averages
  summarize(across(starts_with('num'), mean), .groups = 'drop') %>%
  mutate(grp = glue("{num_sat} Total / {num_big_sat} Big")) %>% 
  ggplot(aes(x = factor(y), y = factor(m), fill = grp)) + 
  geom_tile() + 
  scale_fill_viridis_d(option = "D") + 
  labs(x = "Year of Wedding Date",
       y = "Month of Wedding Date",
       title = "Looking at # Saturdays / ***'BIG'*** Saturdays",
       fill = "") + 
  cowplot::theme_cowplot() + 
  theme(plot.title = ggtext::element_markdown()) +
  coord_flip()

There appears to be a reproducible pattern to which of the three groups you’ll wind up in based on the initial wedding date. Probably not surprisingly this occurs in a 4 year cycle.

• 6 Total / 1 Big – Starts in March after a leap year and continues for the next 12 months.
  • Examples: Mar 2012-Feb 2013, Mar 2016-Feb 2017, Mar 2020-Feb 2021
• 7 Total / 1 Big – The following 12 months after the first group
  • Examples: Mar 2013-Feb 2014, Mar 2017-Feb 2018, Mar 2021-Feb 2022
• 7 Total / 2 Big – The following 24 months after the second group
  • Examples: Mar 2014-Feb 2016, Mar 2018-Feb 2020, Mar 2022-Feb 2024