commute-full.R

  library(googlesheets)
  library(lubridate)
  library(ggplot2)
  library(stringr)

  # authorize the package
  gs_auth()

  # import
  ifttt = gs_title('IFTTT Location Tracker')
  locations = ifttt %>% gs_read(ws = 'Sheet1')

  # de-authorize the package
  gs_deauth()	

  # tidy location name
  home = 'MASKED-FOR-PRIVACY'
  work = 'MASKED-FOR-PRIVACY
  locations = locations %>%
    mutate(site = ifelse(location == home, 'Home', ifelse(location == work, 'Work', 'Other'))) %>%
    select(-location)

  # format date/time
  locations = locations %>%
    mutate(`date/time` = str_replace(`date/time`, 'at ', '')) %>%
    mutate(date_time = mdy_hm(`date/time`)) %>%
    select(-`date/time`)

  # calculate commute times
  n1 = 1
  n2 = 1
  home_to_work_times = list()
  work_to_home_times = list()
  for (i in 1:nrow(locations)) {
    # home to work
    if ((locations$type[i] == 'exited' & locations$site[i] == 'Home') &
        (locations$type[i+1] == 'entered' & locations$site[i+1] == 'Work')) {
      home_to_work_time = difftime(locations$date_time[i+1], locations$date_time[i], units="hours")
      home_to_work_times[[n1]] = c(locations$type[i], locations$site[i], locations$date_time[i], 
                                  locations$type[i+1], locations$site[i+1], locations$date_time[i+1],
                                  'Home to Work', home_to_work_time)
      n1 = n1 + 1
    }

    # work to home
    if ((locations$type[i] == 'exited' & locations$site[i] == 'Work') &
        (locations$type[i+1] == 'entered' & locations$site[i+1] == 'Home')) {
      work_to_home_time = difftime(locations$date_time[i+1], locations$date_time[i], units="hours")
      work_to_home_times[[n2]] = c(locations$type[i], locations$site[i], locations$date_time[i], 
                                  locations$type[i+1], locations$site[i+1], locations$date_time[i+1],
                                  'Work to Home', work_to_home_time)
    }
    n2 = n2 + 1
  }

  # combine, define data types
  commute = bind_rows(
    do.call(rbind, home_to_work_times) %>% as_data_frame(),
    do.call(rbind, work_to_home_times) %>% as_data_frame()
  ) %>% select(start_type = V1, start_place = V2, start_time = V3,
               end_type = V4, end_place = V5, end_time = V6,
               route = V7, hours = V8) %>%
    mutate(hours = as.numeric(hours)) %>%
    mutate(start_time = as.POSIXct(as.numeric(start_time) + 60*60*2, origin = '1970-01-01')) %>%
    mutate(end_time = as.POSIXct(as.numeric(end_time) + 60*60*2, origin = '1970-01-01'))

  # density plot
  plot = commute %>%
    select(Hours = hours, Route = route) %>%
    filter(Hours < 2) %>%
    ggplot(., aes(x = Hours, fill = Route)) +
    geom_density(alpha = .55) +
    theme_minimal() +
    theme(legend.position = 'bottom') + 
    labs(y = 'Density', title = 'Distribution of Commute Times',
         subtitle = 'April - May 2019')

  # export
  setwd("~/Documents/Python/ifttt")
  png('density-plot.png', width = 8, height = 6, units = 'in', res = 600)
  plot
  dev.off()