Introduction to RUGtools

Justin M. Shea

June 13th, 2019

Introduction

R User Group Introduction Slides template

This is a traditional ioslides R Markdown template, but modified to contain default content routinely used when introducing Chicago R user group meetups. Slides are useful because they look good and you won’t forget to do important things like thanking the sponsors! Slides can be accessed from within R Studio using the New R Markdown dialog menu, and then selecting From Template. One can also use the draft function, exemplified below.

rmarkdown::draft(file="new_meetup.Rmd", template = "intro_slides", package="RUGtools")

Estimate your pizza order

In Chicago, we think a lot about Pizza. And if one is involved in the local meetup culture, this is doubly so. At a recent meetup group that wasn’t ours, I counted nearly 6 large pizzas left over. Struck by an overwhelming sorrow, I vowed the Chicago R User Group shall never partake in such a tragic waste of resources. With a few data points, one can use the pizza_estimate function to arrive at a more efficient order.

kable( pizza_estimate(registered = 120, pizza_diameter = 18, attend_rate = 0.57, 
                      serving = 2, style = "thin") )
registered est_attend eaters_per_pizza style pizza_estimate
120 69 5.342811 thin 13

Channeling our ever-curious pizza scientist, it turns out the Chicago “party cut” (thin-crust cut into small squares) inherits a few very attractive properties when dividing p pizzas among n guests. Small square pieces allow guests to better estimate pizza consumption, thus decreasing the integer-programming problem exacerbated by large triangular slices. Reducing wasted pizza is not only virtuous, it demonstrates great stewardship of sponsor resources bestowed upon thee.

Data Analysis

Chicago R User Group data is included and downloaded from meetup.com/ChicagoRUG. Personally identifiable information has been removed, data formatted and ready for analysis.

Load the member list data. How many members do we have?

nrow(member_list)
## [1] 3606

Lets plot the cumulative membership.

plot(y = member_list$Member.ID, x = member_list$Joined.Group.on, type = "l", lwd=2, 
     col = "blue", frame = FALSE, main = "Chicago R User Group members", 
     ylab = "Members", xlab = "", ylim = c(0, 4000), 
     xlim=c(min(member_list$Joined.Group.on), as.Date("2020-01-01")))

How many members joined since January 2017?

nrow(subset(member_list, Joined.Group.on > "2017-01-01"))
## [1] 1554

What percentage of the Chicago R User Group joined since January 2017?

Percentage <- 100 * NROW(subset(member_list, Joined.Group.on > "2017-01-01")) / NROW(member_list)
round(Percentage, 2)
## [1] 43.09

How many new members usually join between meetups?

First, get a unique ordered list of Meetup dates

Meetup_dates <- sort(unique(member_list$Last.Attended))

new_members <- subset(member_list, Joined.Group.on > Meetup_dates[NROW(Meetup_dates)])

Then count the number of new members joined between the most recent meetup and the one prior to that.

new_members2 <- subset(member_list, Joined.Group.on <= Meetup_dates[NROW(Meetup_dates)] &
                                    Joined.Group.on > Meetup_dates[NROW(Meetup_dates)-1])

nrow(new_members2)
## [1] 67

In danger of repeating the above analysis several times over, we created a function new_mem_counter to count the number of new members joined between meetups for all meetups in the data set.

Date New
2010-05-27 0
2010-08-26 20
2010-10-20 10
2010-12-16 15
2011-03-23 34
2011-06-02 17 
kable(tail(new_members), align = 'l')
Date New
65 2018-10-16 101
66 2018-11-14 52
67 2019-01-23 65
68 2019-02-27 50
69 2019-03-20 19
70 2019-05-15 66

Which gap between meetups had the most new members?

kable(new_members[new_members$New==max(new_members$New), ], align = 'l')
Date New
37 2016-06-05 157

Note the previous meetup was 6 months prior, so likely this was not due to the topic covered.

What is the average number of new members joined between meetups?

mean(new_members$New)
## [1] 49.54286

Plot the new members data.

   # Create Date Range Index
  Date_Index <- as.numeric(row.names(new_members[new_members$Date > "2010-01-01" & new_members$Date <= Sys.Date(),]))
  
  # Create x-axis labels, using year-month date format
  x_labels <- format(new_members$Date[Date_Index], "%Y-%m")

  # Plot
  barplot(new_members$New[Date_Index], names.arg = x_labels, main = "CRUG members, joined between meetups",  
          ylab = "New Members", xlab = "")

Plot the new members data since 2017.

   # Create Date Range Index
  Date_Index <- as.numeric(row.names(new_members[new_members$Date > "2017-01-01" & new_members$Date <= Sys.Date(),]))
  
  x_labels <- format(new_members$Date[Date_Index], "%Y-%m")

  # Plot
  barplot(new_members$New[Date_Index], names.arg = x_labels, las=2, main = "CRUG members, joined between meetups",  
          ylab = "New Members", xlab = "")

What is the average number of new members joined between meetups since 2017?

mean(new_members$New[Date_Index])
## [1] 64.33333

Membership as time series

Load and use the xts package.

library(xts)

members_xts <- xts(x = member_list$Member.ID, order.by = member_list$Joined.Group.on)
names(members_xts) <- "useRs"

plot(members_xts, col = "blue", grid.col = "white", main = "Chicago R User Group members by join date")

How many members join by month?

members_monthly <- to.monthly(members_xts, OHLC = FALSE)

barplot(diff(members_monthly), col = "lightgrey", main = "Chicago R User Group, new members by month")

What’s our percentage growth per month?

Omit the first two months growth outliers.

barplot(100*diff(members_monthly)[-c(1,2)]/members_monthly[-c(1,2)], col = "lightgrey",
        main = "CRUG Growth since inception", ylab="Percent")

Let’s view the last 4 years.

members_3_years <- 100*diff(members_monthly["2015-05/"]) / members_monthly["2015-05/"]
barplot(members_3_years, col = "lightgrey", main = "CRUG Growth, last 4 years", ylab="Percent")

Consider seasonal variation.

month_percent_growth <- c(NA, NA, NA, NA, 100*diff(log(coredata(members_monthly))), NA, NA, NA, NA, NA, NA, NA)
seasonal_matrix <- matrix(month_percent_growth, ncol = 12, byrow = TRUE)
colnames(seasonal_matrix) <- month.abb
rownames(seasonal_matrix) <- 2010:2019
seasonal_matrix <- rbind(seasonal_matrix, Median=round(apply(seasonal_matrix, 2, median, na.rm=TRUE), 2))

kable(seasonal_matrix, digits=2, caption = "Percentage Growth per Month")
Percentage Growth per Month
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
2010 NA NA NA NA 32.54 5.41 8.41 17.69 6.54 8.49 2.30 17.66
2011 8.22 5.96 11.69 2.18 8.28 3.90 3.75 6.53 7.21 2.64 8.48 3.76
2012 7.54 8.99 9.67 8.82 13.07 5.54 5.25 7.39 6.66 8.31 7.29 2.81
2013 5.92 3.42 3.15 5.55 5.55 2.59 3.91 6.74 4.13 2.18 1.43 1.18
2014 1.97 4.70 2.91 4.37 5.06 1.82 1.97 2.30 2.25 3.67 3.54 1.56
2015 4.15 3.23 3.42 2.53 2.33 2.14 2.16 2.51 1.04 2.80 2.23 1.88
2016 2.20 1.75 0.92 1.37 1.18 2.59 2.95 2.56 2.14 1.35 0.99 1.03
2017 3.21 2.38 2.86 1.38 2.18 2.35 1.92 1.68 2.21 2.98 2.67 0.60
2018 2.15 2.35 2.93 3.25 2.33 1.47 3.00 1.47 1.20 2.30 1.57 0.50
2019 1.42 1.12 0.74 1.21 0.95 NA NA NA NA NA NA NA
Median 3.21 3.23 2.93 2.53 3.69 2.59 3.00 2.56 2.25 2.80 2.30 1.56

As one of the largest and oldest R user groups in existence, the Chicago R User Group has matured into a comfortable period of value. Growth rates are lower by percentage, but the group continues to serve a steady group of new useRs.