# Class 8: Yedioth Newsvendor Model (📰)
*Aug 1, 2025 | [Transcript](./🦦/8nv(📰)15778_intro_ops_otter_ai.md)*
## 🎯 Teaching Arc
**Hook**: "14,000 SKUs, weekly delivery, unpredictable demand"
**Puzzle**: How optimize inventory when demand is censored?
**Resolution**: Pool data + newsvendor model with service levels
**Model**: Q* = √(2DF/H) + safety stock for uncertainty
## 🗣️ Quality Participation
**Joe** ⭐: "Can update Q after first cycle vs online learning"
- Prof: Appreciated complexity awareness
- Impact: ⭐️"Changing Q complicates policy, makes fragile"
**Key Discussion**: Service level trade-offs
- u/(u+o) formula for target service (95%)
- "Hate runout AND hate inventory"
## 📊 Quick Scores
Focus on students who:
- Question model assumptions
- Identify implementation challenges
- Connect theory to practical constraints
## 🔗 421 Diagram
### Yedioth Model
```
🟢 Weekly delivery 🟣 Small retailers
Sales data 14,000 SKUs
\ /
Coordinate ←→ Compel
/ \
🟠 Min returns 🔴 95% availability
Reduce waste One-stop shop
```
## 📚 Exam Essentials
**Concepts**:
- Newsvendor with censored demand
- Service level: u/(u+o) where u=underage, o=overage
- Lead time uncertainty (DDLT as random variable)
**EOQ Components**:
- D: Demand rate
- F: Fixed order cost ($500)
- H: Holding cost ($0.45)
- C: Unit cost ($45)
**Strategy**: Pool similar products for better demand estimates
**Traps**:
❌"Optimize each SKU separately"
✅"Pool data across similar items"
## 🎓 Recitation Points
- ⭐️"Updating Q frequently makes system fragile"
- "5-day lead time adds complexity"
- "Balance stockout cost vs holding cost"
---
*Previous: [Zara Continued](./7👗_Zara_Continued.md) | Next: [Risk Pooling](./9🏊♀️🚚_Pool_Walmart_Amazon.md)*