S02 / Retail and e-commerce
A merchandise planner must balance excess inventory against missed sales across stores and categories. Ask whether coherent probabilistic forecasts produce more reliable replenishment decisions than independent point forecasts.
Evaluated public-data study
Read the finding and evidence ↗Methods & source
Hierarchical quantile forecasting; newsvendor; stochastic optimization
M5 Walmart sales ↗
Research code ↗
S03 / Retail and e-commerce
An e-commerce team needs to distinguish temporarily inactive buyers from customers unlikely to return. Estimate future transaction counts and revenue distributions to support the timing and prioritization of win-back campaigns.
Evaluated public-data study
Read the finding and evidence ↗Methods & source
BG/NBD; Gamma-Gamma; count/spend challenger
Online Retail II ↗
Research code ↗
S04 / Advertising and marketing technology
A marketing leader wants to reach people whose behavior changes because of advertising. Test whether treatment-effect targeting outperforms response-probability targeting at the same contact capacity.
Evaluated public-data study
Read the finding and evidence ↗Methods & source
Potential outcomes; doubly robust learning; causal forest
Criteo Uplift ↗
Research code ↗
S13 / Manufacturing
A manufacturing quality leader must identify high-risk units without flooding inspection capacity. Test whether a stable, compact sensor model can match a more complex classifier while controlling missed failures.
Evaluated public-data study
Read the finding and evidence ↗Methods & source
Elastic net; PCA process monitoring; stability selection
SECOM ↗
Research code ↗
S28 / Banking and fintech
A sales operations manager has a limited outbound-contact budget. Determine whether calibrated subscription-response models improve the concentration of successful contacts relative to simple recency and prior-contact rules.
Evaluated public-data study
Read the finding and evidence ↗Methods & source
Calibrated classification; additive models; expected-value ranking
Bank Marketing ↗
Research code ↗
S31 / Insurance
An insurance analytics team needs accurate expected-loss estimates that remain understandable across policy segments. Test whether a nonlinear model improves frequency-severity predictions while preserving portfolio and segment calibration.
Evaluated public-data study
Read the finding and evidence ↗Methods & source
Exposure-offset GLMs; frequency-severity; Tweedie models
freMTPL2 frequency and severity ↗
Research code ↗
S43 / Real estate and property management
A real-estate analytics team must distinguish predictable property value from uncertainty caused by sparse local comparables. Test whether flexible spatial models improve sale-price estimates without hiding neighborhood-level error.
Evaluated public-data study
Read the finding and evidence ↗Methods & source
Hedonic/spatial modeling; partial pooling; calibrated intervals
Cook County Assessor parcel sales and characteristics ↗
Research code ↗
S47 / Telecommunications
A telecom service leader wants to understand whether call failures, complaints and usage patterns identify customers at risk before departure. Test stable, interpretable signals and the limits of translating them into an intervention plan.
Evaluated public-data study
Read the finding and evidence ↗Methods & source
Additive models; calibration; causal diagrams; feature stability
Iranian Churn ↗
Research code ↗
S58 / Professional services and enterprise AI
An operations leader needs to know which process paths predict long completion times and where limited review capacity might help. Study bottlenecks and remaining-time uncertainty in a real loan-application workflow.
Evaluated public-data study
Read the finding and evidence ↗Methods & source
Process mining; survival/sequence models; queueing scenarios
BPI Challenge 2017 ↗
Research code ↗
DEMONSTRATION / MARKETING · EXPERIMENTATION
A randomized campaign study connects incremental contribution, uncertainty and the cost of a decision. Evaluated on synthetic customers with known ground truth.
Evaluated synthetic demonstration
Read the finding and evidence ↗Methods & evidence
Causal inference
Synthetic design and reproducible source ↗
DEMONSTRATION / SALES · REVENUE OPERATIONS
A point-in-time, competing-risk forecast estimates when open opportunities will become booked revenue—and exposes the risk of a changing sales process.
Evaluated synthetic demonstration
Read the finding and evidence ↗Methods & evidence
Forecasting · Optimization & simulation
Synthetic design and reproducible source ↗
DEMONSTRATION / CUSTOMERS · RETENTION & VALUE
A contractual survival model and randomized retention study distinguish who may leave, what a customer is worth, and whether paying to intervene creates value.
Evaluated synthetic demonstration
Read the finding and evidence ↗Methods & evidence
Survival analysis · Causal inference
Synthetic design and reproducible source ↗
DEMONSTRATION / OPERATIONS · DEMAND & CAPACITY
A weekly staffing study joins demand distributions, exact capacity constraints and overtime decisions to measure the cost–service tradeoff on future demand.
Evaluated synthetic demonstration
Read the finding and evidence ↗Methods & evidence
Forecasting · Optimization & simulation
Synthetic design and reproducible source ↗