Add multi-step define/add funcs

[yl3394]

Can we have more than one step data-manipulation functions? For example

data %>%   group_by(age_range, 
           d_imp_rate_level) %>%
  summarise(n_users_mean = round(mean(n_users))) %>% 
  ungroup() %>%
  group_by(age_range) %>%
  mutate(total_n_users_mean = sum(n_users_mean),
         pctg_users_mean = percent(n_users_mean/total_n_users_mean, accuracy = 0.01))

The second step computation is based on the the output of step 1.

[teramonagi]

Using our example data, you can convert the following code

df <- dummy_data
library("dplyr")
df %>%
  group_by(gender, age) %>%
  summarise(impression = mean(impression)) %>% 
  ungroup() %>%
  group_by(gender) %>%
  mutate(total_impression = sum(impression), pctg_impression = impression/total_impression)

with

metrics_1 <- mmetrics::define(impression = mean(impression))
metrics_2 <- mmetrics::define(total_impression = sum(impression), pctg_impression = impression/total_impression)
df %>%
  mmetrics::add(gender, age, metrics = metrics_1) %>%
  mmetrics::add(gender, metrics = metrics_2, summarize=FALSE)