.webp)
Table of Contents
Key takeaways
- Predictions is a new AI-enabled Pigment feature that generates data-driven forecasts with advanced statistical and machine learning models without requiring data science expertise.
- These forecasts combine historical data with external drivers to predict future values for demand planning, financial forecasting, and revenue projections.
- Predictions aims to save users significant time while providing a more accurate view into the future of the organization.
- The Prophet model fits forecasting scenarios with at least two years of historical data, clear seasonality, and factors such as holidays, special events, marketing spend, or promotions.
- AutoETS fits cases with less than two years of historical data and simpler patterns with clear, repetitive trends.
- Seasonal differencing fits situations where recent data matters more than older history and seasonal highs and lows recur at similar times each year.
- After selecting a model and defining parameters, including external factors, the forecast is generated and can be published in Pigment.
Imagine you’re preparing for your annual budget cycle. It’s never a fun time.
The CFO is asking for a revenue forecast for next year that’s more accurate than what we’ve prepared before.
You have three years of actuals available, but there are so many variables at play (pricing updates, marketing spend, etc) that your normal manual approach to forecasting just isn’t going to cut it.
Now, there’s a solution available within just a few clicks.
Introducing Predictions
Predictions is a new AI-enabled feature that allows users to generate data-driven forecasts using advanced statistical and machine learning models - and you don’t need any data science expertise to do so.
These models analyze historical data alongside external drivers to predict future values, and are useful for a wide range of planning use cases including demand planning, financial forecasting, and revenue projections.
It’s going to save users a significant amount of time and allow you a more accurate window into the future of your organization.
Selecting the right model
Predictions includes a few different models you can forecast with. Depending on your use case, you’ll want to employ a specific model.
- Prophet
Suitable when you have at least two years of historical data, with clear seasonal patterns, and when you may need to account for holidays, special events or external drivers (marketing spend, promotions, etc.) - AutoETS
Suitable when you have less than two years of historical data. Great for simpler data patterns with clear, repetitive trends. - Seasonal differencing
Suitable when recent data is more relevant than older historical data, and when seasonal highs and lows happen around the same time every year.
Once you’ve selected a model and defined parameters - like which external factors to include in - your forecast is generated and can be published.
Start forecasting now
Predictions is available in Pigment today. To learn more about how it works and how to gain access, visit the Pigment Community.
