Capacity Prediction in External Services
summary
Predicts vendor capacity by modeling utilization as a linear function of a load factor = Σ(request_rate × duration) and fits that model in Microsoft Excel, yielding R² = 0.85. Contract cap is 650 executable units and the model flags exceeding it at a load factor ≈ 29,000. Validated against a real sale where load jumped from ~5,000 to ~25,000 and predictions closely match observed utilization. Uses 95th percentile durations in a slider spreadsheet and plans a live feed feeding ML for automation.
AI-written summary · about
chapters (13)
- 0:00Intro
- 1:50How the airline industry works
- 2:49Data flow
- 3:47Shopping API
- 4:30Request Types
- 6:23Request Rates
- 8:12Request Duration
- 8:41ITA Reports
- 11:55Model
- 13:43Load Factors
- 14:13Validate Model
- 21:08Next Steps
- 21:52Questions
related talks
21m"Capacity Prediction" instead of "Capacity Planning":
29mCapacity Planning with the Universal Scalability Law
53mScale your business in real-time using ML in Microsoft Fabric Real-time Analytics
59mPredictability Check: Demand Forecasting
48mLoad Testing Fabric - What is our hold capacity?
28mDemand Forecasting: Accuracy, Stability, Automation
20mManaging Capacity
52mSQL Server Predictive Analytics: Customer Stories