All about Advanced Predictions in Oracle EPM.

Advanced Predictions – Why This Feature Matters

Forecasting in the real world is rarely driven by a single number. Revenue is influenced by pricing, promotions, volume, market growth, seasonality, macro‑economic conditions, and many other factors. Until now, Oracle EPM’s Auto Predict and Predictive Planning features looked at only one measure at a time, which limited how realistic and explainable forecasts could be.

With the August 2025 Oracle Cloud EPM release, Oracle introduces Advanced Predictions – a significant shift toward driver‑based, machine‑learning forecasting that feels native to finance teams, not data scientists.

Advanced Predictions allows planners to:

• Use multiple internal and external drivers
• Apply enterprise‑grade ML algorithms (automatically)
• Understand why a forecast looks the way it does
• Stay entirely inside the EPM Planning experience

This blog walks through the feature slowly, visually, and practically – the way you’d explain it to a business user, a solution architect, or a finance leader.

 

1. What Exactly Is Advanced Predictions?

At a high level, Advanced Predictions brings multivariate forecasting into Oracle EPM Planning and FreeForm.

Simple way to explain it

• Auto Predict answers:

“Based on past revenue trends, what might future revenue look like?”

• Advanced Predictions answers:

“Based on revenue, pricing changes, promotions, industry growth, and macro drivers – what is the most likely future outcome, and why?”

Behind the scenes, Oracle uses OCI Data Science Cloud and modern ML algorithms, but the user experience remains finance‑friendly and UI‑driven.

 

Core Differences at a Glance

 

Aspect	Auto Predict / Predictive Planning	Advanced Predictions
Forecast approach	Univariate	Multivariate
Drivers	Single measure	Multiple drivers
ML engine	Statistical	OCI ML algorithms
Algorithm choice	Fixed	LightGBM, XGBoost, Prophet, SARIMAX, AutoMLx
Explainability	Limited	Rich, interactive explanations

 

2. Business Problems This Solves

Advanced Predictions is not about replacing planners – it’s about augmenting judgment with intelligence.

Real‑world planning challenges

• Revenue forecasts ignore pricing changes
• Cost forecasts miss inflation or volume impacts
• Manual driver models are hard to maintain
• Forecasts lack transparency and trust

 

What Advanced Predictions changes

• Forecasts reflect multiple business realities
• Planners can see driver impact, not just results
• ML complexity is hidden, but insight is exposed
• Finance teams stay inside EPM – no data science tools needed

 

Common use cases

• Revenue forecasting using price, volume, promotions, GDP
• Expense forecasting using volume, inflation indices
• Sales planning using industry growth + internal drivers

 

3. How Advanced Predictions Works

Although the technology is sophisticated, the process is easy to explain.

End‑to‑end flow

1. EPM prepares historical data and driver series
2. OCI Data Science trains ML models
3. Multiple algorithms are evaluated
4. The best‑performing model is selected
5. Forecasts and accuracy metrics are written back to EPM

 Important reassurance for IT & Security

• No separate OCI subscription required
• No custom code
• Oracle fully manages orchestration

 

4. Enabling Advanced Predictions (Administrator View)

Advanced Predictions is opt‑in, which means administrators control when and how it becomes available.

IPM Configuration Wizard home screen

 

What admins do here

• Enable Advanced Predictions
• Define where it appears
• Control exposure to planners

5. Machine Learning Algorithms 

One of the most impressive parts of Advanced Predictions is that users don’t need to understand ML to benefit from it.

Supported algorithms

Algorithm	What it’s good at	When it shines
LightGBM	Complex relationships	Large structured datasets
XGBoost	High accuracy	Non‑linear driver impacts
Prophet	Seasonality	Business time‑series
SARIMAX	Statistical + drivers	Econometric‑style forecasting
AutoMLx	Automatic selection	Most business scenarios

AutoMLx runs multiple models and automatically picks the best result, which is ideal for finance teams.

 

6. Planner Experience – Where the Value Shows Up

Once enabled, planners interact with Advanced Predictions through dashboards and forms, not technical tools.

Overview Dashboard

Planners can immediately see:

• Forecasted trends
• Drivers used

• Scenario outcomes

 

Advanced Prediction overview dashboard

Deep‑dive analysis views

View	What planners learn
Prediction	Forecast vs history
Input Drivers	Driver values & influence
Forecast Accuracy	Model reliability
Predict by Algorithms	Side‑by‑side model comparison

This visibility builds trust, which is critical for adoption.

 

7. Explain Prediction – The “Aha” Feature

This is where Advanced Predictions truly feels different.

How planners use it

1. Open a form with predicted values
2. Right‑click a predicted cell
3. Select Explain Prediction

What they see

• Historical trend
• Best case / worst case / most likely scenarios
• Visual explanation of outcomes

This answers the most common user question:

“Why does this number look like this?”

 

8. Key Things You Should Know (Before You Enable)

•  Available for Planning & FreeForm
•  Supports BSO and ASO cubes
•  Web UI only (Explain Prediction not in Smart View)
•  Requires deliberate enablement by admins

 

📋 Quick Checklist Table

 

Area	Ready?
Historical data quality	⬜
Driver availability	⬜
Planner training	⬜
Governance model	⬜

 

9. When Should You Use (or Avoid) Advanced Predictions?

Use it when:

• Forecast accuracy truly matters
• Business outcomes depend on multiple drivers
• You want explainable forecasting

Maybe skip it when:

• Simple trend projection is enough
• Data history is too sparse
• Drivers are not reliable

Advanced Predictions augments planning, it doesn’t replace business judgment.

 

Advanced Predictions marks a clear shift in Oracle EPM:

➡ From spreadsheet‑style forecasting

➡ To intelligent, driver‑aware, explainable planning

It brings machine learning to finance users, on their terms, without technical friction. For many organizations, this will be the first time AI forecasting actually feels usable – and trustworthy. 

Happy days on the Cloud!!!