IL-REC
Ilmek Öneri
Real-time personalised product, content and campaign recommendations for e-commerce, telecom and finance; churn…
Forecasting & Recommendation View
Ilmek Tahmin
Forecasting & Recommendation
IL-FCT
Hierarchical time-series forecasting for demand, sales, stock and energy load, with probabilistic ranges and automatic explanations.
- Probabilistic forecast ranges
- Cold-start forecasting
- Campaign impact analysis
- Explained forecasts
Accurate forecasting lowers stock cost and prevents lost sales. Ilmek Tahmin models thousands of time series at once across hierarchies such as product–store–region, and automatically considers external factors such as calendar, campaigns, weather and price.
Capabilities
- Probabilistic forecasting: confidence intervals (P10 / P50 / P90) instead of a single mean
- Cold-start forecasts for new products based on similar items
- Campaign impact and price elasticity analysis
- Writing forecasts into planning systems (ERP, WMS) and deviation alerts
- An explanation showing the driving factors for each forecast
For retail and energy customers, forecast error (WAPE) has fallen 25–35% and stock turnover has risen.
How to start
A backtest is run with 2–3 years of historical data and compared with your current forecasting method. Results are presented by product category and store, and the pilot proceeds in the areas with the biggest gap.
| Hierarchy | Product × store × region |
|---|---|
| Output | P10 / P50 / P90 |
| Horizon | 1 day – 18 months |
| Integration | ERP, WMS, data warehouse |