f@0: ---
f@0: title: "Analyse Sets"
f@0: output: html_notebook
f@0: ---
f@0: 
f@0: ```{r}
f@0: source('../db/access_db.R')
f@0: library(tidyverse)
f@0: ```
f@0: 
f@0: 
f@0: ```{r}
f@0: excerpts_classes <- get_excerpts_classes()
f@0: 
f@0: excerpts_artist <- get_excerpts_artists()
f@0: 
f@0: classes_artists <- excerpts_classes %>%
f@0:   inner_join(excerpts_artist, by = c('ex_id')) %>%
f@0:   inner_join(get_class_names(), by = c('class_id'))
f@0: 
f@0: ```
f@0: 
f@0: ## Excerpts
f@0: 
f@0: ```{r}
f@0: sets_excerpts <- rbind(
f@0:   read.csv('../sets/train_fixed.csv') %>% mutate(set = 'train'),
f@0:   read.csv('../sets/test_fixed.csv') %>% mutate(set = 'test'),
f@0:   read.csv('../sets/filt_fixed.csv') %>% mutate(set = 'filt')) %>%
f@0:   inner_join(excerpts_classes, by = c('ex_id'))
f@0: 
f@0: (sets_excerpts_table <- 
f@0:   sets_excerpts %>% 
f@0:   group_by(class_id, set, iter) %>%
f@0:   unique() %>%
f@0:   summarise(num_excerpts = n()) %>%
f@0:   ungroup())
f@0: ```
f@0: 
f@0: 
f@0: ```{r }
f@0: 
f@0: sets_excerpts_table$set <- factor(sets_excerpts_table$set,
f@0:                                   levels = c('train', 'test', 'filt'))
f@0: 
f@0: sets_excerpts_table$set <- plyr::mapvalues(sets_excerpts_table$set,
f@0:                                   from = c('filt', 'test', 'train'),
f@0:                                   to = c('Pr. Test',
f@0:                                          'Test',
f@0:                                          'Train'))
f@0: sets_excerpts_table <- sets_excerpts_table %>%
f@0:   inner_join(get_class_names(), by = c('class_id'))
f@0: ```
f@0: 
f@0: ```{r fig.height = 3, fig.width = 6}
f@0: excerpts_p <- ggplot(sets_excerpts_table) +
f@0:   geom_violin(aes(x = set,
f@0:                    y = num_excerpts,
f@0:                    color = set, fill = set),
f@0:               alpha = 0.6, size = 1,
f@0:               draw_quantiles = c(0.5)) +
f@0:   theme_bw() +
f@0:   scale_y_continuous(name = 'Unique Excerpts',
f@0:                      limits=c(0, 80)) +
f@0:   xlab('Collection') +
f@0:   theme(axis.title.x = element_text(size = 12),
f@0:         axis.title.y = element_text(size = 11),
f@0:         axis.text = element_text(size = 11),
f@0:         axis.text.x = element_text(
f@0:           angle = 45, hjust = 1, vjust = 1),
f@0:         legend.position = 'none', 
f@0:         strip.text.x = element_text(size = 11)) +
f@0:   facet_grid(.~class)
f@0: ```
f@0: 
f@0: ## Artists
f@0: 
f@0: ```{r}
f@0: sets_artists <- sets_excerpts %>%
f@0:   inner_join(excerpts_artist, by = c('ex_id'))
f@0: 
f@0: (sets_artists_table <- 
f@0:   sets_artists %>% 
f@0:   group_by(class_id, set, iter) %>%
f@0:   summarise(num_artists = n_distinct(artist_id)) %>%
f@0:   ungroup())
f@0: ```
f@0: 
f@0: 
f@0: ```{r }
f@0: 
f@0: sets_artists_table$set <- factor(sets_artists_table$set,
f@0:                                   levels = c('train', 'test', 'filt'))
f@0: 
f@0: sets_artists_table$set <- plyr::mapvalues(sets_artists_table$set,
f@0:                                   from = c('filt', 'test', 'train'),
f@0:                                   to = c('Pr. Test',
f@0:                                          'Test',
f@0:                                          'Train'))
f@0: sets_artists_table <- sets_artists_table %>%
f@0:   inner_join(get_class_names(), by = c('class_id'))
f@0: ```
f@0: 
f@0: ```{r fig.height = 3, fig.width = 6}
f@0: artists_p <- ggplot(sets_artists_table) +
f@0:   geom_violin(aes(x = set,
f@0:                    y = num_artists,
f@0:                    color = set, fill = set),
f@0:               alpha = 0.6, size = 1,
f@0:               draw_quantiles = c(0.5)) +
f@0:   theme_bw() +
f@0:   scale_y_continuous(name = 'Unique Artists',
f@0:                      limits=c(0, 60)) +
f@0:   xlab('Collection') +
f@0:   theme(axis.title.x = element_text(size = 12),
f@0:         axis.title.y = element_text(size = 11),
f@0:         axis.text = element_text(size = 11),
f@0:         axis.text.x = element_text(
f@0:           angle = 45, hjust = 1, vjust = 1),
f@0:         legend.position = 'none', 
f@0:         strip.text.x = element_text(size = 11)) +
f@0:   facet_grid(.~class)
f@0: ```