load(here::here("/data/influence_network.RData"))
disruptions = read_csv(here::here("data/artist-disruptions_summarized.csv"))

## Parsed with column specification:
## cols(
##   id = col_character(),
##   name = col_character(),
##   disruption_mean = col_double(),
##   confidence = col_double()
## )

degrees = read_csv(here::here("data/artist-network-degrees.csv"))

## Parsed with column specification:
## cols(
##   label = col_character(),
##   name = col_character(),
##   earliest_decade = col_double(),
##   genre = col_character(),
##   indegree = col_double(),
##   outdegree = col_double()
## )

disruptions = disruptions %>% 
    left_join(degrees, by = c("id" = "label", "name"))

skimr::skim(disruptions)

## Skim summary statistics
##  n obs: 14287 
##  n variables: 8 
## 
## ── Variable type:character ───
##  variable missing complete     n min max empty n_unique
##     genre     137    14150 14287   3  14     0       20
##        id       0    14287 14287  12  12     0    14287
##      name       0    14287 14287   1  61     0    14129
## 
## ── Variable type:numeric ─────
##         variable missing complete     n    mean    sd      p0       p25
##       confidence       0    14287 14287    0.8   0.12    0.49    0.75  
##  disruption_mean       0    14287 14287    0.17  0.19   -0.33    0.0015
##  earliest_decade    1553    12734 14287 1971.14 22.09 1890    1960     
##         indegree       0    14287 14287    8.4  29.81    1       1     
##        outdegree       0    14287 14287    3.91  5.5     0       0     
##      p50     p75    p100     hist
##     0.75    0.88    1    ▁▁▁▂▇▁▂▃
##     0.24    0.25    0.98 ▁▁▇▇▂▁▁▁
##  1980    1990    2010    ▁▁▁▃▃▇▆▂
##     2       5    1492    ▇▁▁▁▁▁▁▁
##     1       7      36    ▇▂▁▁▁▁▁▁

disruptions %>% 
    filter(outdegree > 0) %>% 
    ggplot(aes(x = confidence, y = disruption_mean)) + 
    geom_point(size = .8, alpha = .4, color = "#F58A07") 

disruptions %>% 
    filter(outdegree > 0) %>% 
    ggplot(aes(x = confidence)) + 
    geom_density(fill = "salmon")

disruptions %>% 
    filter(confidence > .9) %>% 
    ggplot(aes(x = disruption_mean)) +
    stat_ecdf(geom = "line") + 
    labs(
        x = "D", 
        y = "P(x < D)"
    )

disruptions %>% 
    filter(confidence >= .9, outdegree > 0) %>% 
    ggplot(aes(x = disruption_mean)) +
    geom_density(fill = "#F58A07", alpha = .6) + 
    labs(
        x = "D", 
        y = "Density" 
    )

disruptions %>% 
    filter(confidence >= .9, outdegree > 3) %>% 
    summarise(
        cor_d_i = cor(disruption_mean, indegree, method = "kendall"), 
        cor_d_o = cor(disruption_mean, outdegree, method = "kendall")
    )

## # A tibble: 1 x 2
##     cor_d_i cor_d_o
##       <dbl>   <dbl>
## 1 -0.000544  -0.198

disruptions %>% 
    filter(outdegree > 3) %>% 
    ggplot(aes(x = earliest_decade, y = confidence, group = earliest_decade)) + 
    geom_boxplot(coef = 1e3)

## Warning: Removed 25 rows containing missing values (stat_boxplot).

disruptions %>% 
    filter(confidence >= .9, outdegree >= 3) %>% 
    ggplot(aes(x = earliest_decade, y = disruption_mean, group = earliest_decade)) + 
    geom_boxplot(coef = 1e3) + 
    geom_jitter(width = .9, alpha = .4, color = "salmon", size = .7)

## Warning: Removed 12 rows containing missing values (stat_boxplot).

## Warning: Removed 12 rows containing missing values (geom_point).

disruptions %>%
    mutate(abs_disruption = abs(disruption_mean)) %>%
    filter(confidence >= .9,
           indegree >= 3,
           outdegree >= 3,
           abs_disruption > .05,
           earliest_decade >= 1950) %>%
    ggplot(aes(
        x = indegree,
        y = outdegree,
        size = abs_disruption,
        color = disruption_mean
    )) +
    geom_point(alpha = .6) +
    scale_x_log10() +
    scale_y_log10() +
    scale_color_gradient2(
        low = "#122338",
        mid = "#344065",
        high = "#FF96F6",
        midpoint = 0
    )

disruptions %>% 
    filter(confidence >= .9, outdegree >= 3, earliest_decade >= 1930) %>% 
    arrange(desc(disruption_mean)) %>% 
    slice(1:20) %>% 
    kable()

disruptions %>% 
    filter(confidence >= .9, outdegree >= 10, earliest_decade >= 1930) %>% 
    arrange(desc(disruption_mean)) %>% 
    slice(1:20) %>% 
    kable()

disruptions %>%
    filter(confidence >= .9,
           outdegree >= 3,
           indegree >= 200,
           earliest_decade >= 1950) %>%
    arrange(desc(disruption_mean)) %>%
    slice(1:20) %>% 
    kable()

disruptions %>% 
    filter(confidence >= .9, outdegree >= 3, earliest_decade >= 1950) %>% 
    arrange((disruption_mean)) %>% 
    slice(1:20) %>% 
    kable()

disruptions %>% 
    filter(confidence >= .9, outdegree >= 10, earliest_decade >= 1950) %>% 
    arrange((disruption_mean)) %>% 
    slice(1:20) %>% 
    kable()

disruptions %>%
    filter(confidence >= .9,
           outdegree >= 3,
           indegree >= 100,
           earliest_decade >= 1950) %>%
    arrange(disruption_mean) %>%
    slice(1:20) %>% 
    kable()

visualize_influences = function(allmusic_id, network, node_info) {
    create_focal_network(allmusic_id, network, node_info) %>%
        ggraph(layout = "kk") +
        # geom_node_point(aes(colour = type, shape = order), size = 4) +
        geom_edge_link(aes(
            start_cap = label_rect(node1.name),
            end_cap = label_rect(node2.name)
        ),
        arrow = arrow(length = unit(2, 'mm'))) +
        geom_node_label(aes(label = name, color = type)) +
        # geom_node_label(aes(label = name)) +
        theme_void() +
        NULL

}

# visualize_influences("mn0000022845", influences, nodes)
# create_focal_network("mn0000022845", influences, nodes) %>%
#     activate()
# 
# create_focal_network("mn0000764636", influences, nodes) %>%
#     activate(nodes) %>%
#     as_tibble() %>% View
# 
# create_focal_network("mn0000754032", influences, nodes) %>%
#     activate(nodes) %>%
#     as_tibble() %>% View
# 
# visualize_influences("mn0001052511", influences, nodes)
# 
# create_focal_network("mn0000052151", influences, nodes) %>%
#     activate(nodes) %>%
#     as_tibble() %>% View