Use cases
From a decision problem to measured evidence
Each use case defines the decision, data source, evaluation method, measured result, and limitation. The industry structure separates common methods from market-specific assumptions.
Industries
Four industries, five applied cases
2 use cases
Energy
Price forecasts and dispatch signals for flexible energy assets.
View industry
1 use case
Maritime
Vessel arrival forecasts for port capacity and service planning.
View industry
1 use case
Retail
Demand forecasts across products, stores, and planning levels.
View industry
1 use case
Advertisement
Causal response curves for budget allocation across channels and markets.
View industryCase library
Methods, results, and stated limits
Forecast market prices before battery dispatch
A shared forecasting method estimates day-ahead and reserve-market prices, then converts each price curve into a charge and discharge schedule for battery assets.
The case estimates about EUR 4.5 million of additional annual day-ahead value per GWh of battery capacity for an illustrative German two-hour asset.
Read use case
Re-price flexible power at every market gate
A rolling forecast produces a 24-hour price curve at every half-hour gate, which updates the signal as the delivery period approaches.
Across 194 delivery days, the rolling signal adds SGD 33,000 per MW relative to a previous-day schedule under the stated spread-value assumptions.
Read use caseEvaluate vessel ETA forecasts across planning horizons
An ongoing engagement evaluates vessel ETA forecasts across the planning horizons that support port capacity and service decisions.
The engagement remains in progress, so the underlying data and results cannot be disclosed at this stage.
Read use caseForecast demand across 30,490 product-store series
The case uses the public Walmart M5 forecasting benchmark dataset to compare a coherent 28-day demand forecast across 30,490 product-store series.
The case records a WRMSSE of 0.738, compared with 0.765 for the traditional benchmark, and completes a nationwide forecast in about three minutes.
Read use caseReallocate ad spend with causal response curves
A causal media-mix method accounts for demand effects and estimates marginal return by channel and market. It assigns a fixed budget to the highest estimated marginal returns.
The case estimates 13% more attributed revenue at the same total budget, with a 90% interval from 12% to 29%.
Read use caseEvaluate a representative workload with eomer.
A 30-minute session covers baseline results, uncertainty, deployment options, and the limits of production use.
Request demo