Taking Your Statistical Forecast to the Next Level

Innovation Blog Series: Forecast Engine Tuning

Every industry vertical is unique, each with its own market and market driving conditions. This inherently leads to its own set of forecasting-related constraints. In addition, every business within a given industry vertical has its own flavor of the forecasting process. The “out-of-the-box” settings of the forecasting engine may not provide the best baseline forecast for the demand planning team. There is a need for forecast accuracy improvement which can be met by tailoring the engine settings to a business’s specific forecasting needs. This is commonly known as Engine Tuning. The challenge is to identify the avenues for forecast accuracy improvement and to determine the engine settings which will provide the desired results.

Improving Customer Service Levels while Lowering Inventory Carrying Costs

Based on our experience working with more than a hundred customers from various industry verticals, we have developed a proven methodology for tuning the statistical engine. The process has three key steps:

  1. Demand Data Analysis (DA) – Data analysis identifies demand patterns in history and classifies them into five types: Stable, Volatile, Intermittent, Too Few, and Others. The distribution of combinations and their historical quantities in these buckets helps us identify the nature of demand, providing valuable inputs for optimal engine settings. Additionally, this information provides excellent insights to the business demand planning team in identifying combinations that are easy to forecast vs. difficult to forecast.
  2. Engine Settings Changes – The statistical forecast could change significantly for the same input (history) based on changes to Engine Settings. These changes include a combination of one more or more of 4 attributes, namely, forecast tree reconfiguration, causality modeling, forecasting methods selection, and engine parameter changes.
  3. Advanced Engine Tuning using Analysis Sets – In addition to the system-wide settings changes, we can also perform advanced engine tuning, which starts with creating Analysis Sets (demand segmentation) and then creating forecasting profiles specific to these Analysis Sets. Data analysis results are used to create multiple analysis sets and fine-tune the engine based on particular demand patterns. This approach is similar to the followed for nodal tuning exercise in the on-premise Oracle Demantra® solution.

 

Accuracy Improvements for Oracle Demand Management Cloud Customers

At each of the above steps, we collaborate closely with the business demand planning team to ensure success. We perform quantitative (detailed data over short planning horizons, every three to six months) and qualitative analysis (graphs over long term planning horizons, one to three years) of the statistical forecast. The qualitative guidance/observations by the business team play a crucial role in the holistic tuning of the forecast.

Using this methodology, we have successfully achieved significant accuracy improvements for our customers using Oracle Demand Management Cloud® and Oracle Demantra® in diverse industry verticals such as High Tech, Industrial Manufacturing, CPG, Retail, Life sciences, Oil and Gas, and Food & Beverages, to name a few.
 

 

Inspirage has the answers to your questions

A tuned forecasting engine provides a better baseline statistical forecast at the beginning of the Demand Planning process, the benefits of which cascade to all the downstream processes such as Supply Planning, Inventory, Manufacturing, Procurement of the Supply Chain. Inspirage’s Forecast Engine Tuning delivers the accuracy needed meet your business’s specific forecasting needs.

If you are on the fence about moving to Oracle Cloud due to gaps between your business requirements and Oracle’s Cloud offering, it is possible that we already built a solution to bridge that gap. If we do not have a prebuilt solution to address your challenges, we can customize something to fit. We will do the work behind the scenes, so you can focus on growing your business. Please visit the Oracle Cloud Marketplace to see a complete listing of Inspirage Solutions and contact us with any questions or to schedule a demo.

Kaushik Dinkar Mantri | Key Contributor

Kaushik Dinkar Mantri is a Manager at Inspirage with over 16 years of work experience in Technology Consulting, Information Technology, and Supply Chain Management. His areas of expertise include Supply Chain Analytics, Demand Data Modeling, and Data Analysis, focusing on Forecasting, Demand Planning, and Inventory Management. He has rich experience working with clients across the globe in Manufacturing, Life Sciences, High Tech, and Food & Beverages industry verticals, among others. Kaushik is a certified Lean Six Sigma professional. He is also a certified Oracle Inventory Management specialist with a master's degree in Industrial Engineering from Purdue University.