Tracing the Rise of Szymon Pyć and Predicting the Next Breakout Star

Author

Zhang Xuerong

Published

June 1, 2025

OVERVIEW

Introduction

In this report, we analyze the rise of artist Szymon Pyć using a social network perspective and knowledge graph data provided in Mini-Challenge 1 of the VAST Challenge 2025. The objective is twofold:

  1. Trace the professional rise of Szymon Pyć.
  2. Predict the next breakout artist using centrality and influence analysis.

Methodology

We employ a visual analytics and social network analysis framework. This involves constructing and analyzing a graph of artists, albums, and collaborations, and using metrics such as Eigenvector Centrality to detect influential nodes.

DATA PREPARATION

Installing and Loading Packages

Loading the Raw Dataset

  • json_data <- jsonlite::fromJSON("/Users/sharon/OneDrive - Singapore Management University/isss608data/take-home_exercise2/MC1_graph.json")

# Load and transform nodes and edges together safely
nodes <- as_tibble(json_data$nodes) %>%
  mutate(
    name = name,  # Keep igraph name field
    display_name = ifelse(is.na(name), as.character(id), name),
    label = make.unique(as.character(name))
  )


edges <- as_tibble(json_data$links)

Data Wrangling

# Identify Szymon Pyć's ID
szymon_id <- nodes %>% filter(display_name == "Szymon Pyć") %>% pull(id)

# Get relevant edges and nodes
szymon_edges <- edges %>% filter(source == szymon_id | target == szymon_id)
connected_ids <- union(szymon_edges$source, szymon_edges$target)
szymon_works <- nodes %>% 
  filter(id %in% connected_ids, `Node Type` %in% c("Song", "Album")) %>%
  mutate(release_year = as.numeric(release_date),
         notoriety_year = as.numeric(notoriety_date))

VISUALISATION AND ANALYSIS

szymon_summary <- szymon_works %>%
  group_by(release_year) %>%
  summarise(
    total_releases = n(),
    notable_releases = sum(notable, na.rm = TRUE)
  )

Release Timeline

ggplot(szymon_summary, aes(x = release_year)) +
  geom_col(aes(y = total_releases), fill = "steelblue") +
  geom_line(aes(y = notable_releases), color = "darkred", size = 1.2) +
  labs(title = "Szymon Pyć's Release Timeline", y = "Releases", x = "Year") +
  theme_minimal()
Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
ℹ Please use `linewidth` instead.

Insight: Szymon’s notable release activity peaked in 2003, suggesting a breakthrough year. Follow-up years had fewer releases, hinting at a concentrated period of fame.

Genre Distribution Over Time

genre_trend <- szymon_works %>%
  filter(!is.na(genre)) %>%
  group_by(release_year, genre) %>%
  summarise(count = n())
`summarise()` has grouped output by 'release_year'. You can override using the
`.groups` argument.
ggplot(genre_trend, aes(x = release_year, y = count, fill = genre)) +
  geom_area() +
  theme_minimal() +
  labs(title = "Genre Distribution of Szymon Pyć's Work", y = "Count")

Insight: This reveals genre shifts and experimentation. Peaks in new genres may signal strategic attempts to reach wider audiences.

Graph Network Creation and Centrality

# Clean and preserve full node attributes
nodes_clean <- nodes %>%
  filter(!is.na(id), id != "", id != "NA") %>%
  distinct(id, .keep_all = TRUE) %>%
  mutate(id = as.character(id))

# Confirm presence of Node Type
glimpse(nodes_clean)
Rows: 17,412
Columns: 12
$ `Node Type`    <chr> "Song", "Person", "Person", "Person", "RecordLabel", "S…
$ name           <chr> "Breaking These Chains", "Carlos Duffy", "Min Qin", "Xi…
$ single         <lgl> TRUE, NA, NA, NA, NA, FALSE, NA, NA, NA, NA, TRUE, NA, …
$ release_date   <chr> "2017", NA, NA, NA, NA, "2026", NA, NA, NA, NA, "2020",…
$ genre          <chr> "Oceanus Folk", NA, NA, NA, NA, "Lo-Fi Electronica", NA…
$ notable        <lgl> TRUE, NA, NA, NA, NA, TRUE, NA, NA, NA, NA, TRUE, NA, N…
$ id             <chr> "0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "10",…
$ written_date   <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, "2020", NA, NA,…
$ stage_name     <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,…
$ notoriety_date <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,…
$ display_name   <chr> "Breaking These Chains", "Carlos Duffy", "Min Qin", "Xi…
$ label          <chr> "Breaking These Chains", "Carlos Duffy", "Min Qin", "Xi…
# Revalidate edge structure
edges_clean <- edges %>%
  mutate(source = as.character(source), target = as.character(target)) %>%
  filter(source %in% nodes_clean$id & target %in% nodes_clean$id) %>%
  filter(source != target) %>%
  distinct(source, target, .keep_all = TRUE)

glimpse(edges_clean)
Rows: 31,714
Columns: 4
$ `Edge Type` <chr> "InterpolatesFrom", "RecordedBy", "PerformerOf", "Composer…
$ source      <chr> "0", "0", "1", "1", "2", "2", "3", "5", "5", "5", "5", "5"…
$ target      <chr> "1841", "4", "0", "16180", "0", "16180", "0", "5088", "143…
$ key         <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
# Build graph with extra validation
library(igraph)

edges_clean <- edges %>%
  mutate(source = as.character(source), target = as.character(target)) %>%
  filter(source %in% nodes_clean$id & target %in% nodes_clean$id) %>%
  filter(!is.na(source), !is.na(target),
         source != "", target != "",
         source != "NA", target != "NA") %>%
  filter(source != target) %>%
  distinct(source, target, .keep_all = TRUE)

head(edges_clean)
# A tibble: 6 × 4
  `Edge Type`      source target   key
  <chr>            <chr>  <chr>  <int>
1 InterpolatesFrom 0      1841       0
2 RecordedBy       0      4          0
3 PerformerOf      1      0          0
4 ComposerOf       1      16180      0
5 PerformerOf      2      0          0
6 ProducerOf       2      16180      0
stopifnot(all(edges_clean$source %in% nodes_clean$id))
stopifnot(all(edges_clean$target %in% nodes_clean$id))

graph_tbl_full <- tbl_graph(nodes = nodes_clean, edges = edges_clean[, c("source", "target")], node_key = "id", directed = TRUE)

head(graph_tbl_full)
6 x 17412 sparse Matrix of class "dgCMatrix"
  [[ suppressing 34 column names 'Breaking These Chains', 'Carlos Duffy', 'Min Qin' ... ]]
                                                                               
Breaking These Chains . . . . 1 . . . . . . . . . . . . . . . . . . . . . . . .
Carlos Duffy          1 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Min Qin               1 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Xiuying Xie           1 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Nautical Mile Records . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Unshackled Heart      . . . . . . . . . 1 . . . . . . . . . . . . . . . . . . .
                                      
Breaking These Chains . . . . . ......
Carlos Duffy          . . . . . ......
Min Qin               . . . . . ......
Xiuying Xie           . . . . . ......
Nautical Mile Records . . . . . ......
Unshackled Heart      . . . . . ......

 .....suppressing 17378 columns in show(); maybe adjust options(max.print=, width=)
 ..............................
# Extract largest weakly connected component
components <- components(as.igraph(graph_tbl_full), mode = "weak")
main_component_ids <- V(as.igraph(graph_tbl_full))$name[components$membership == which.max(components$csize)]

graph_tbl <- graph_tbl_full %>%
  filter(name %in% main_component_ids) %>%
  mutate(eigen_centrality = centrality_eigen())

head(graph_tbl)
6 x 17279 sparse Matrix of class "dgCMatrix"
  [[ suppressing 34 column names 'Breaking These Chains', 'Carlos Duffy', 'Min Qin' ... ]]
                                                                               
Breaking These Chains . . . . 1 . . . . . . . . . . . . . . . . . . . . . . . .
Carlos Duffy          1 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Min Qin               1 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Xiuying Xie           1 . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Nautical Mile Records . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Unshackled Heart      . . . . . . . . . 1 . . . . . . . . . . . . . . . . . . .
                                      
Breaking These Chains . . . . . ......
Carlos Duffy          . . . . . ......
Min Qin               . . . . . ......
Xiuying Xie           . . . . . ......
Nautical Mile Records . . . . . ......
Unshackled Heart      . . . . . ......

 .....suppressing 17245 columns in show(); maybe adjust options(max.print=, width=)
 ..............................
# Top influencers
top_influencers <- graph_tbl %>% 
  as_tibble() %>% 
  filter(`Node Type` == "Person") %>% 
  arrange(desc(eigen_centrality)) %>% 
  slice(1:50)
top_influencers %>% select(display_name, eigen_centrality)
# A tibble: 50 × 2
   display_name     eigen_centrality
   <chr>                       <dbl>
 1 Kimberly Snyder             1    
 2 Gerald Mullins              0.633
 3 Joshua Herring              0.560
 4 Kelly Stewart               0.400
 5 John Anderson               0.315
 6 Rachel Jackson              0.293
 7 Roy Tschentscher            0.239
 8 Erica Walton                0.178
 9 Angelina Sölzer             0.173
10 Thomas Weber                0.170
# ℹ 40 more rows
# Visualize top 5% most central nodes
top_threshold <- quantile(graph_tbl %>% as_tibble() %>% pull(eigen_centrality), 0.95, na.rm = TRUE)
top_subgraph <- graph_tbl %>% filter(eigen_centrality >= top_threshold)

graph_central_plot <- ggraph(top_subgraph, layout = "fr") +
  geom_edge_link(alpha = 0.2) +
  geom_node_point(aes(size = eigen_centrality, color = `Node Type`)) +
  geom_node_text(aes(label = display_name), repel = TRUE, size = 2.5) +
  theme_void() +
  labs(title = "Top 5% Most Influential Nodes")

print(graph_central_plot)
Warning: ggrepel: 862 unlabeled data points (too many overlaps). Consider
increasing max.overlaps

Insight: High eigenvector scores reflect influence in the music graph. These individuals often collaborate or get referenced frequently.

Visualization of Szymon Pyć’s Ego Network

szymon_ego <- graph_tbl_full %>% convert(to_subgraph, name %in% connected_ids)
Subsetting by nodes
# DIAGNOSTICS
cat("Szymon Ego Node Count:", nrow(as_tibble(szymon_ego)), "
")
Szymon Ego Node Count: 0 
szymon_ego <- graph_tbl %>% convert(to_subgraph, id %in% connected_ids)
Subsetting by nodes
graph_plot <- ggraph(szymon_ego, layout = "fr") +
  geom_edge_link(alpha = 0.3) +
  geom_node_point(aes(color = `Node Type`), size = 3) +
  geom_node_text(aes(label = display_name), repel = TRUE, size = 2.5) +
  theme_void() +
  labs(title = "Szymon Pyć's Collaboration and Influence Network")

print(graph_plot)

PREDICTING THE NEXT BREAKOUT STAR

# Check which node types are present in the graph
graph_tbl %>%
  as_tibble() %>%
  count(`Node Type`)
# A tibble: 5 × 2
  `Node Type`      n
  <chr>        <int>
1 Album          988
2 MusicalGroup   220
3 Person       11292
4 RecordLabel   1196
5 Song          3583
# DIAGNOSTICS
top_new_df <- graph_tbl %>%
  as_tibble() %>%
  filter(`Node Type` == "Person") %>%
  arrange(desc(eigen_centrality)) %>%
  slice(1:10)

head(top_new_df)
# A tibble: 6 × 13
  `Node Type` name          single release_date genre notable id    written_date
  <chr>       <chr>         <lgl>  <chr>        <chr> <lgl>   <chr> <chr>       
1 Person      Kimberly Sny… NA     <NA>         <NA>  NA      1716  <NA>        
2 Person      Gerald Mulli… NA     <NA>         <NA>  NA      1719  <NA>        
3 Person      Joshua Herri… NA     <NA>         <NA>  NA      1718  <NA>        
4 Person      Kelly Stewart NA     <NA>         <NA>  NA      1918  <NA>        
5 Person      John Anderson NA     <NA>         <NA>  NA      8250  <NA>        
6 Person      Rachel Jacks… NA     <NA>         <NA>  NA      8248  <NA>        
# ℹ 5 more variables: stage_name <chr>, notoriety_date <chr>,
#   display_name <chr>, label <chr>, eigen_centrality <dbl>
summary(top_new_df$eigen_centrality)
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
 0.1699  0.1934  0.3042  0.3962  0.5199  1.0000 
ggplot(top_new_df, aes(x = reorder(display_name, eigen_centrality), y = eigen_centrality)) +
  geom_col(fill = "tomato") +
  coord_flip() +
  labs(title = "Top Potential Breakout Artists by Eigenvector Centrality",
       x = "Artist", y = "Eigenvector Score") +
  theme_minimal()

Deep Dive: Breakout Candidate — Kimberly Snyder

k_snyder_id <- nodes %>% filter(display_name == "Kimberly Snyder") %>% pull(id)
k_snyder_edges <- edges %>% filter(source == k_snyder_id | target == k_snyder_id)
k_snyder_ids <- union(k_snyder_edges$source, k_snyder_edges$target)
k_snyder_works <- nodes %>% 
  filter(id %in% k_snyder_ids, `Node Type` %in% c("Song", "Album")) %>%
  mutate(release_year = as.numeric(release_date),
         notoriety_year = as.numeric(notoriety_date))

ggplot(k_snyder_works, aes(x = release_year, fill = notable)) +
  geom_bar() +
  labs(title = "Kimberly Snyder's Yearly Output and Notable Work",
       x = "Release Year", y = "Number of Works") +
  theme_minimal()

Insight: Kimberly Snyder shows strong recent output and rising notoriety. Her central position and prolific work closely mirror Szymon Pyć’s pre-breakout phase.

LEARNING POINTS

  • Temporal analysis highlights Szymon’s breakthrough year (2003).
  • Genre transitions reflect stylistic evolution.
  • Network metrics help identify up-and-coming stars.
  • Social graph visualization links influence and collaboration.
  • Interactive elements add engagement and exploration.
  • Breakout candidate analysis offers grounded prediction.