R Notebook

Exploring user behavior on Snapshots at Sea

library(data.table)
library(dplyr)
library(dtplyr)
library(ggplot2)
library(treemap)
library(stringr)
library(magrittr)

Define functions

#summarize input file: group by user and count classifications 
summarize_user_classifications <- function(dataset, col_name){
     dataset$user <- dataset[[col_name]]
     data_out <- dataset %>% 
          mutate(., user = ifelse(str_detect(user, pattern = "not-logged-in"), "not-logged-in", user)) %>%
          group_by(., user) %>%
          summarise(., num_classifications = n()) %>% #count the number of classifications per user
          arrange(., -num_classifications) %>% #don't know why desc(num_classifications) isn't working
          add_rownames() #index for the treemap
     return(data_out)
}

Read in and summarize user data: in this case, snapshots at sea

classifications <- fread("../Data/snapshots-at-sea-classifications.csv", select = 1:8)

Read 0.0% of 1085163 rows
Read 14.7% of 1085163 rows
Read 28.6% of 1085163 rows
Read 41.5% of 1085163 rows
Read 56.2% of 1085163 rows
Read 71.0% of 1085163 rows
Read 83.9% of 1085163 rows
Read 94.9% of 1085163 rows
Read 1085163 rows and 8 (of 14) columns from 1.150 GB file in 00:00:14

demo_dat <- summarize_user_classifications(classifications, "user_name")

Deprecated, use tibble::rownames_to_column() instead.

print(demo_dat)

Look at user classification activity

ggplot(data = dat, aes(x=num_classifications)) + geom_density()

ggplot(data = dat, aes(x=num_classifications)) + geom_density() + scale_x_log10()

ggplot(data = dat, aes(x=num_classifications)) + geom_histogram(bins = 100) + scale_x_log10()

Looking at the distribution for just the lowest 100 classifiers

ggplot(data = filter(dat, num_classifications <= 100), aes(x=num_classifications)) + geom_bar() 

ggplot(data = filter(dat, num_classifications <= 25), aes(x=num_classifications)) + geom_bar() 

treemap(dtf = dat, index = "rowname", vSize = "num_classifications", border.lwds = .1, fontsize.labels = 4)