📉 Customer Churn Predictor
Predict and prevent policy cancellations
Model Description
Predicts the likelihood of customers canceling their insurance policies, enabling proactive retention efforts.
Churn Risk Factors
| Factor | Weight | Description |
|---|---|---|
| Payment Issues | High | Late/missed payments |
| Claim Denied | High | Recent claim rejection |
| Price Increase | Medium | Premium went up |
| Life Event | Medium | Moving, marriage, etc. |
| Competitor Offer | Medium | Shopping for quotes |
| Low Engagement | Low | No app/portal usage |
| Policy Age | Low | New customers higher risk |
Risk Segments
| Segment | Churn Probability | Action |
|---|---|---|
| 🟢 Loyal | <10% | Maintain relationship |
| 🟡 At Risk | 10-30% | Proactive outreach |
| 🟠 High Risk | 30-60% | Retention offer |
| 🔴 Critical | >60% | Immediate intervention |
Performance
| Metric | Score |
|---|---|
| AUC-ROC | 0.89 |
| Precision | 84% |
| Recall | 81% |
| Lift @10% | 4.2x |
Business Impact
- 15% reduction in voluntary churn
- $2.3M saved annually in retention
- 23% improvement in customer lifetime value
Usage
import pandas as pd
from transformers import pipeline
predictor = pipeline("tabular-classification", model="gcc-insurance-ml-models/customer-churn-predictor")
customer = {
"tenure_months": 8,
"claims_count": 1,
"claim_denied": True,
"payment_issues": 0,
"premium_increase_pct": 12,
"app_logins_30d": 0,
"support_tickets": 2
}
result = predictor(customer)
# Output: {'churn_probability': 0.67, 'segment': 'critical', 'top_factors': ['claim_denied', 'premium_increase']}
Retention Playbook
Churn Score > 60%
↓
[Identify Top Factors]
↓
claim_denied → Escalate review, offer goodwill
price_increase → Discount offer, bundle savings
payment_issues → Flexible payment plan
low_engagement → Personalized outreach
License
Apache 2.0