Fabric Planning: Build Statistical Forecasts using ETS and ARIMA models

In today’s volatile market, enterprise agility relies heavily on the accuracy of demand projections. While historical data tells us where we have been, advanced statistical forecasting tells us where we are going. 

Join us for this deep-dive session where we explore how to architect and execute sophisticated statistical forecasts directly within the Microsoft Fabric ecosystem.  

This webinar will bypass generic black-box solutions to focus on a comprehensive "Built from Scratch" approach—demonstrating how to leverage your existing Semantic Models, enrich them, and generate robust future projections using the native, inbuilt Forecasting Engine of the Fabric Plan work item. 

What We Will Cover 

We will break down the mechanics and implementation paths of two powerhouse forecasting methodologies: 

• Error, Trend, Seasonal (ETS) Frameworks: Moving from foundational Single Exponential Smoothing to Holt’s Model (Double), and into complex Holt-Winter’s Models (Triple)—including a look at how to apply Additive, Multiplicative, and Damped extensions to capture nuanced seasonal behaviors.
   

• Autoregressive Integrated Moving Average (ARIMA) Models: Mastering both Non-Seasonal ARIMA (covering Autoregressive [AR] models, Differenced AR, and Random Walk behaviors) and Seasonal ARIMA (SARIMA) variations, including General SARIMA, Seasonal AR, and the classic Airline Model. 

Key Takeaways 

• Learn how to configure the Fabric Plan Forecasting Engine. 

• Discover how to bridge the gap between a standard BI Semantic Model and an active, predictive EPM (Enterprise Performance Management) environment.