FORECAST ENGINE DEDICATED TO THE RETAIL BUSINESS
In order to fill the constant need for up to date information of retailers, Soft Solutions embeds within the Retail Business Suite a forecast engine. ibs Analytics is a service dedicated to information modeling and simulations, owned by Soft Solutions. It optimizes the latest modeling and information mining techniques in a retail context in order to fully map the science on the business and operational needs.
“The science at the service of the retail”
In today retail context, the understanding of the market evolution and clients needs is a central part of the business strategy to "drive" the retailers in everyday decision making for any kind of business: offer management, promotion planning, pricing, ordering and supply chain management. With the intelligence provided by the ibs Analytics embedded forecast engine, retailers can access an additional level of understanding in a transparent manner and take benefits of it directly in the Retail Business Suite.
Since version 6 of the Soft Solutions Retail Business Suite, the ADDM (Advanced Data Decomposition Model) approach has been implemented to model and forecast sales and stocks at the level of the items / stores / days. This approach has proven its efficiency and accuracy in production contexts, showing the main advantages of the ADDM:
- Accurate simulation results for use at different execution levels (stores, categories…)
- Modular approach allowing fast implementation and ROI
- Optimized approach allowing to deal with Tier 1 scopes (500,000+ items, 5,000+ stores…) real time
- "Vulgarized"science to decompose results for business users understanding and demystify the science aspects. This last point, coupled with the ibs AMS service actually pushes the science at the level of the business to guarantee that end users will take full benefits of the ibs Analytics forecast engine.
The various implementations have driven our solution to include various kinds of models and configurations to ensure most
accurate results to our client. These ones are defined and managed by our research laboratory for the clients
with proven results.
Despite these results and in order to continue enhancing science for the business, ibs Analytics has evolved to a meta-forecast engine solution. This approach allows ibs Analytics engine to adjust automatically to any situation. Depending on the items / stores, it defines automatically the required models within a large library and apply them to provide the most accurate forecast simulations.
“A better accuracy with less resources”
The engine can then reserve complex models (time consuming) to harder simulations, saving time on the simpler ones (very stable level of sales for example / non seasonal). As a result, the engine automatically adapts to the changes in the business and ensures to keep constant optimal level of accuracy in forecast simulations. We also noticed an important earn in calculation time in parallel to an increase in overall accuracy of the forecasts.
- Meta-engine approach to apply the most efficient model to any item / store.
- Forecast models library
- Items sales patterns clustering (seasonal, cyclic, permanent…)
Advanced Data Decomposition Model
- Advanced Data Decomposition Model
- Non parametric and signal processing techniques tuned to match retail context
- Service-based approach to include additional indicators (elasticity, cannibalization…)
Definition & Analysis of The Information
- Outliers detection based on non-statistical techniques
- Outliers correction based on neighborhood analysis
Today with the meta-forecast engine approach, ibs Analytics takes full benefits of its original ADDM forecast models that have proven their efficiency and maximizes overall accuracy by including an additional layer, authorizing to individually define the best models for each item.