helpq: Interface to the 6.004 queue

# library(pool)


#
# q_add_series <- function(dta) {
#   function(hc, dim, ...) {
#     hc %>%
#       hc_add_series_times_values(dates  = dta[["time"]],
#                                  values = dta[[dim]],
#                                  step   = TRUE,
#                                  ...    = ...)
#   }
# }
#
# shinyServer(function(input, output) {
#   ## Data ======================================================================
#   date_range <- reactive({
#     c(start = input$date_range[1] - days(1) + hours(3),
#       end   = input$date_range[2] + days(0) + hours(3))
#   })
#

#
#   wait_data <- reactive({
#     wait_data <- student_q() %>%
#       mutate(t_id = dense_rank(time)) %>%
#       filter(!is.na(username)) %>%
#       group_by(username) %>%
#         arrange(time) %>%
#         mutate(req_id = (t_id - lag(t_id, default = -Inf) > 1) |
#                  (!being_helped & lag(being_helped, default = !first(being_helped))),
#                req_id = cumsum(req_id)) %>%
#       group_by(username, req_id) %>%
#         mutate(start_wait = min(time[!being_helped]),
#                end_wait   = max(time[!being_helped]),
#                start_help = min(time[ being_helped]),
#                end_help   = max(time[ being_helped]),
#                t_wait     = time - start_wait,
#                wait       = end_wait - start_wait) %>%
#         filter(time %in% c(start_wait, end_wait, start_help, end_help)) %>%
#         ungroup()
#
#     units(wait_data$wait) <- "mins"
#     units(wait_data$t_wait) <- "mins"
#
#     return(wait_data)
#   })
#
#   current_q <- reactive({
#     student_q() %>% top_n(1, time)
#   })
#
#   ## Top row ===================================================================
#   output$n_staff <- renderInfoBox({
#     infoBox(value = current_q()[["n_staff"]][[1]],
#             title = "staff in lab",
#             icon  = icon("users"),
#             color = "yellow",
#             width = 3)
#   })
#
#   output$n_helping <- renderInfoBox({
#     infoBox(value = sum(current_q()[["being_helped"]]),
#             title = "students being helped",
#             icon  = icon("comments"),
#             color = "yellow",
#             width = 3)
#   })
#
#   output$n_students <- renderInfoBox({
#     infoBox(value = sum(!current_q()[["being_helped"]]),
#             title = "students waiting",
#             icon  = icon("question"),
#             color = "yellow",
#             width = 3)
#   })
#
#   output$wait <- renderInfoBox({
#     wait_last <- wait_data() %>%
#       top_n(1, time) %>%
#       filter(!being_helped)
#
#     if (nrow(wait_last) == 0) {
#       return(ib <- infoBox(value = NA,
#                            title = "wait time",
#                            icon  = icon("clock-o"),
#                            color = "green",
#                            fill  = T,
#                            width = 3))
#     }
#
#     wait_time <- round(max(wait_last[["t_wait"]]), digits = 2)
#     color <- "green"
#     if (wait_time > 15) {
#       color <- "orange"
#     }
#     if (wait_time > 30) {
#       color <- "red"
#     }
#
#     infoBox(value = wait_time,
#             title = "wait time",
#             icon  = icon("clock-o"),
#             color = color,
#             fill  = T,
#             width = 3)
#   })
#
#   ## Plots =====================================================================
#   output$plot_all <- renderHighchart({
#     q_data <- student_q() %>%
#       group_by(time) %>%
#         mutate(staff_username = ifelse(is.na(position), NA, staff_username),
#                n_helping = n_distinct(staff_username, na.rm = TRUE),
#                n_waiting = n_helping - n_distinct(username, na.rm = TRUE)) %>%
#         ungroup() %>%
#       mutate(n_helping_d = n_helping != lag(n_helping),
#              n_waiting_d = n_waiting != lag(n_waiting),
#              n_staff_d   = n_staff   != lag(n_staff)) %>%
#       filter(n_waiting_d | n_helping_d | n_staff_d)
#
#     q_add <- q_add_series(q_data)
#
#     highchart() %>%
#       q_add("n_helping",
#             name  = "Helped",
#             color = "green",
#             type  = "area") %>%
#       q_add("n_waiting",
#             name = "Waiting",
#             color = "red",
#             type  = "area") %>%
#       q_add("n_staff",
#             name  = "Staff",
#             color = "black")
#   })
#
#   output$plot_staff_utilization <- renderHighchart({
#     q_data <- student_q() %>%
#       arrange(time) %>%
#       group_by(time) %>%
#         mutate(staff_utilization = 100*sum(being_helped)/n_staff) %>%
#         ungroup() %>%
#       filter(staff_utilization != lag(staff_utilization))
#
#     q_add <- q_add_series(q_data)
#
#     highchart() %>%
#       q_add("staff_utilization",
#             name = "Percent staff utilized",
#             type = "area") %>%
#       hc_yAxis(min = 0, max = 100)
#   })
#
#   output$plot_staff_view <- renderPlot({
#     staff_q() %>%
#       ggplot(aes(x = time,
#                  y = factor(staff, levels = unique(staff)),
#                  color = is_helping)) +
#       geom_point() +
#       theme_ipsum_rc() +
#       labs(x = NULL, y = NULL)
#   })
#
#   output$plot_wait_time <- renderPlot({
#     ylims <- c(0, max(wait_data()$wait[wait_data()$wait < 120]))
#
#     wait_data() %>%
#       ggplot(aes(x = time,
#                  y = t_wait,
#                  color = being_helped,
#                  alpha = !being_helped,
#                  group = paste(username, req_id))) +
#
#       geom_line(size = .5,
#                 alpha = .85) +
#
#       geom_point(data = wait_data() %>%
#                   group_by(username) %>%
#                   filter(time %in% start_help),
#                 aes(x = time,
#                     y = t_wait,
#                     alpha = T,
#                     group = "1"),
#                 color = "black",
#                 alpha = 1,
#                 size = 1.2) +
#
#       scale_y_continuous(breaks = 0:100*10,
#                          minor_breaks = 0:100*5) +
#       scale_color_discrete(guide = F) +
#       scale_alpha_discrete(guide = F) +
#
#       coord_cartesian(ylim = ylims) +
#       theme_ipsum_rc() +
#       labs(x = NULL, y = NULL)
#   })
#
#   output$q_char <- renderPlot({
#     wd <- student_q() %>%
#       mutate(t_id = dense_rank(time)) %>%
#       filter(!is.na(username)) %>%
#       group_by(username) %>%
#       arrange(time) %>%
#       mutate(req_id = (t_id - lag(t_id, default = -Inf) > 1) |
#                (!being_helped & lag(being_helped, default = !first(being_helped))),
#              req_id = cumsum(req_id)) %>%
#       group_by(username, req_id) %>%
#       mutate(start_wait = min(time[!being_helped]),
#              end_wait   = max(time[!being_helped]),
#              start_help = min(time[ being_helped]),
#              end_help   = max(time[ being_helped]),
#              t_wait     = time - start_wait,
#              wait       = end_wait - start_wait) %>%
#       ungroup()
#
#     units(wd$wait) <- "mins"
#     units(wd$t_wait) <- "mins"
#
#     wd %>%
#       group_by(username, req_id) %>%
#         arrange(time) %>%
#         mutate(w = as.numeric(lead(time, default = max(time)) - time,
#                               units = "mins") *
#                  !being_helped) %>%
#         summarize(succ     = any(being_helped),
#                   n_staff  = weighted.mean(n_staff,w),
#                   wait     = first(wait),
#                   position = max(position_adj)) %>%
#         ungroup() %>%
#       filter(succ,
#              wait < 180,
#              n_staff > 0.2,
#              position > 0) %>%
#
#       ggplot(aes(x = position/n_staff,
#                  y = as.numeric(wait),
#                  color = n_staff,
#                  fill = n_staff)) +
#       geom_point(position = position_jitter(width = 0, height = 0)) +
#       theme_ipsum_rc()
#   })
#
#   output$expect <- renderPlot({
#     wd <- wait_data() %>%
#       filter(time > as.POSIXct("2017-10-26 03:00:00"))
#
#     wd_rate <- wd %>%
#       group_by(username, req_id) %>%
#         top_n(1, time) %>%
#         ungroup() %>%
#       # group_by(staff_username) %>%
#       filter(!is.na(end_help),
#              as.numeric(end_help) > -Inf) %>%
#       mutate(help_t = end_help - start_help) %>%
#       summarize(rate = mean(as.numeric(end_help - start_help, units = "mins"))) %>%
#       .[["rate"]]
#
#     wd %>%
#       ggplot(aes(x = time,
#                  y = t_wait,
#                  color = being_helped,
#                  alpha = !being_helped,
#                  group = paste(username, req_id))) +
#       geom_line(size = .5,
#                 alpha = .85) +
#       geom_point(data = wd %>%
#                   filter(n_staff != 0) %>%
#                   arrange(time) %>%
#                   mutate(t_now = max(time)) %>%
#                   group_by(username, req_id) %>%
#                     filter(any(being_helped) | any(t_now == time)) %>%
#                     summarize(n_staff = first(n_staff),
#                               help_t  = first(start_wait),
#                               q_len   = first(position_adj),
#                               wait_time = first(wait)) %>%
#                     ungroup(),
#                 aes(x = help_t,
#                     y = q_len/n_staff*wd_rate),
#                 inherit.aes = F,
#                 color = "green", alpha = 1) +
#       geom_point(data = wd %>%
#                    group_by(username) %>%
#                    filter(time %in% start_help),
#                  aes(x = time,
#                      y = t_wait,
#                      alpha = T,
#                      group = "1"),
#                  color = "black",
#                  alpha = 1,
#                  size = 1.2) +
#       scale_y_continuous(breaks = 0:100*10,
#                          minor_breaks = 0:100*5) +
#       scale_color_discrete(guide = F) +
#       scale_alpha_discrete(guide = F) +
#       # coord_cartesian(ylim = c(0, max(wd$wait[wd$wait < 50]))) +
#       theme_ipsum_rc() +
#       labs(x = NULL, y = NULL)
#   })
# })