A network-based cultural analytics model tracing how influence travels across five South African genres and eight decades — from township jazz to global Afrotech.
South African music didn't evolve in a straight line — each era absorbed and mutated what came before. The influence didn't disappear. It compounded. Click any era to go deeper.
The model's most fertile periods aren't the middle of each era — they're the overlaps. When Kwaito and House coexisted (1998–2002), when House and Amapiano collided (2012–2016), that's when the network density spikes and the most interesting cross-genre edges form. Transitions aren't endings. They're the most creative moments in the entire graph.
These scores don't measure streams or awards. They measure structural importance — how central an artist was to how the music actually evolved. Illustrative data
Each shape shows how an artist scores across the five influence dimensions. No single artist dominates every axis — and that's the point.
Every transition had a specific artist acting as a sonic translator — someone whose creative language was legible to two different musical communities at once. That's the rarest ability this model can measure.
Afrotech is a producer's genre — Afro House fused with techno's darker energy, African percussion, and predominantly Xhosa and Zulu vocals. It's built in studios, runs 6–9 minutes for club versions, and travels through DJ sets before streaming ever catches up. Focalistic and Uncle Waffles are Amapiano artists — important, correctly placed in that era. The artists below are the ones who actually authored this sound.
Global streaming data built on Johannesburg. It missed Durban's gqom entirely until Dlala Thukzin's numbers made it impossible to ignore. It missed Thakzin creating 3-Step in Ivory Park during lockdown before a single Spotify play existed. The intelligence layer — manual annotation from interviews, festival lineups, club sets, local blog archives — is what separates a data project from a cultural argument.
Every significant artist in the model maps to one of four structural roles. The role tells you how they shaped the music — not just whether they did.
Each artist's Influence Score decomposed across five weighted dimensions. The breakdown explains why an artist scores where they do. In Amapiano, the producers outscore the vocalists — and the model is correct to show that. Illustrative data
| Artist | Archetype | Era | Network | Cross-Genre | Innovation | Longevity | Cultural | IS |
|---|
This project exists to demonstrate a specific way of thinking — that cultural phenomena can be modelled rigorously without reducing them to popularity metrics, and that the most interesting data often lives outside the API.
NetworkX, directed weighted graphs, centrality algorithms — degree, betweenness, eigenvector, PageRank. Temporal subgraph analysis across six overlapping era windows.
Multi-source fusion across Spotify, Discogs, Wikidata, MusicBrainz. Fuzzy deduplication, schema design, ETL in Pandas. Manual annotation pipeline for the intelligence layer.
Qualitative signal encoded as structured data. The model knows that Amapiano spread on WhatsApp before Spotify. That Durban's gqom seeded Afrotech's percussion. APIs don't know that. The researcher does.
Gephi force-directed network graphs. Plotly interactive dashboards. Temporal influence evolution animations. The graph is the argument made visible.
"I modelled the evolution and influence flow of South African music using network-based cultural analytics — quantifying how musical ideas, production techniques, and artist relationships propagate across genres and decades. The data confirmed what the culture already knew."