Multiple layers over time w/ geom_box months

multigeoms.R

if (!require("pacman")) install.packages("pacman"); library(pacman)
p_load(tidyverse, numform, viridis, ggrepel, lexicon)
p_load_current_gh('trinker/plotflow')

set.seed(13)
dat <- data_frame(
    revenue = rnorm(5000, 50000, 5000),
    date = sample(seq(as.Date('1999/01/01'), as.Date('1999/12/31'), by="day"), 5000, TRUE),
    site = sample(paste("Site", 1:5), 5000, TRUE),
    sales_person = sample(f_title(sample(lexicon::common_names, 15)), 5e3, TRUE)
) %>%
    mutate(
        mon = factor(format(date, format = '%b'), levels = constant_months_abbreviation)
    ) %>%
    group_by(mon) %>%
    mutate(
        revenue = {if(sample(0:1, 1) == 0) `-` else `+`}(revenue, sample(1e2:1e4, 1)),
        observe = revenue > 7.05e4 | revenue < 3.0e4
    ) %>%
    ungroup() 

mondat <- data_frame(
    xmin = seq(as.Date('1999/01/01'), as.Date('1999/12/01'), by="month"),
    xmax = seq(as.Date('1999/02/01'), as.Date('2000/01/01'), by="month"),
    mon = factor(format(xmin, format = '%b'), levels = constant_months_abbreviation)
) %>%
    group_by(mon) %>%
    mutate(xmid = mean(c(xmin, xmax)))


sumdat <- dat %>%
    group_by(mon) %>%
    summarize(revenue = mean(revenue)) %>%
    left_join(mondat)

sumdat2 <- dat %>%
    group_by(mon) %>%
    summarize(med = median(revenue), upper = quantile(revenue, .90), lower = quantile(revenue, .10)) %>%
    left_join(mondat)

se <- function(x) sqrt(var(x)/length(x))

sumdat3 <- dat %>%
    group_by(mon) %>%
    summarize(med = mean(revenue), upper = mean(revenue) + 1.96 * se(revenue), lower = mean(revenue) - 1.96 * se(revenue)) %>%
    left_join(mondat)


set.seed(11)
ggplot(dat) +
    geom_rect(data = mondat, 
        aes(xmin = xmin, xmax = xmax, ymin = 2e4, ymax = Inf, fill = mon), 
        alpha = .1
    ) +
    geom_jitter(height = 0, width = 1, aes(x = date, y = revenue, alpha = observe)) +
    geom_text(data = mondat, aes(x = xmid, y = 8.2e4, label = mon)) +
    geom_hline(data = summarize(dat, revenue = mean(revenue)), 
        aes(yintercept = revenue),
        linetype = 'dashed', color = 'grey10', size = .7
    ) +
    #geom_segment(
    #     data = sumdat, 
    #     aes(y = revenue, yend = revenue, x = xmin, xend = xmax),
    #     color = 'gray45', size = .9#, linetype = 'dotted'
    #) +
    geom_pointrange(data = sumdat2, aes(x = xmid, y = med, ymin = lower, ymax = upper)) +
    geom_smooth(aes(x = date, y = revenue), color = 'blue', fill = NA) +
    theme_minimal() +
    scale_y_continuous(labels = numform::ff_denom(prefix = '$')) +
    theme(
        axis.ticks.x = element_blank(), 
        axis.text.x = element_blank(), 
        legend.position = 'none', 
        panel.grid.major.x = element_blank(), 
        panel.grid.minor.x = element_blank()
    ) +
    scale_fill_manual(values = rep(c('grey60', 'white'), 6)) +
    labs(
        y = 'Revenue', x = 'Date'
    ) +
    scale_alpha_discrete(range = c(.05, 1)) +
    geom_text_repel(
        data = subset(dat, observe),
        aes(label = sales_person, x = date, y = revenue),
        size = 3,
        color = 'grey60',
        box.padding = 0.25,
        point.padding = 0.3, 
    ) 




ggsave('multiple_geoms.pdf', width = 12, height = 5)