Geek Genre Text Classifier (Baseline)
Model description
This lightweight baseline classifies short, original geek-fiction descriptions as cyberpunk, high_fantasy, space_opera, or superhero.
It uses a TF-IDF text vectorizer followed by logistic regression, saved as a joblib artifact.
Training data
The model was trained on a small original seed dataset from the companion GitHub project. The texts were written specifically for this educational portfolio project.
Metrics and limitations
The first reproducible run obtained 31.25% accuracy on 16 evaluation examples after training on 48 examples. Metrics are stored in metrics.json and come from a small stratified holdout split. They measure performance only on this limited synthetic seed dataset, not on real fiction, fandom discourse, or production traffic.
Usage
import joblib
model = joblib.load("model.joblib")
model.predict(["A rebel pilot races through an asteroid field to protect a colony."])
License
MIT