Dc-4nderson/cic-skillbridge-models

SkillBridge — Coding in Color Recommendation Models

Two ML models that power skill recommendations for the Coding in Color program.

Architecture

Model 1 (K-Means)  →  student archetype
                          ↘
                           LLM  →  skill rec + project idea
                          ↗
Model 2 (XGBoost)  →  follow-through probability

Model 1: Skill Cluster (K-Means)

• Input: 36 integer skill columns (checkin counts per skill)
• Output: Cluster name, confidence, distances

Model 2: Engagement Predictor (XGBoost)

• Input: 7 engagement features (no skill selection — that's the LLM's job)
• Output: Follow-through probability (0-1)

Usage

import joblib
import xgboost as xgb
from huggingface_hub import hf_hub_download

cluster_model = joblib.load(hf_hub_download("Dc-4nderson/cic-skillbridge-models", "model1/cluster_model.joblib"))
scaler = joblib.load(hf_hub_download("Dc-4nderson/cic-skillbridge-models", "model1/scaler.joblib"))

eng_model = xgb.XGBClassifier()
eng_model.load_model(hf_hub_download("Dc-4nderson/cic-skillbridge-models", "model2/engagement_model.json"))