Identity modeled as dynamic archetypal states, not a fixed type.
ML
NLP
HMM
LSTM
HDBSCAN
LDA
SHAP
Python
//FIG.14 — dissolve
An ML framework that models identity as dynamic archetypal states — Jungian theory met with archetypal analysis, LDA, HDBSCAN, Hidden Markov Models, LSTM, and SHAP explainability, with a prototype "Archetype Navigator" that forecasts transitions.
//Role
Principal investigator
//Status
Research / prototype
//Access
Live / public
//Problem
Personality frameworks freeze people into fixed types. Identity is better understood as movement between states over time — which static models can’t capture.
//Architecture
A pipeline combining archetypal analysis, LDA, and HDBSCAN for structure, Hidden Markov Models and an LSTM for the temporal dynamics, and SHAP for explainability — surfaced through a prototype "Archetype Navigator" that forecasts transitions.
//Role
Principal investigator — the theory, the modeling stack, and the prototype.
//Outcome
A research framework with a working transition-forecasting prototype.
What it took
Technical proof.
▸Models identity as dynamic archetypal states rather than fixed traits.
▸Jungian theory + archetypal analysis, LDA, HDBSCAN, Hidden Markov Models, and LSTM.
▸SHAP explainability over the state-transition model.