Power BI integrated forecast method

[LennartPfeiler]

Hello everyone,

I am currently a student and have the task of optimising a Power BI report forecast. The report just uses the integrated forecast method in the analytics panel. I have already researched extensively and have somehow not found a consistent result as to which model Power BI uses. I have previously seen exponential smoothing, ARIMA, regression, winter holts, etc. Could someone tell me which model it really is? As I also understood, there are 2 different models. Is there any kind of documentation explaining when which algorithm is selected? I would be very grateful, thank you!
Any sources would be amazing, thank you!

[Anonymous]

Hi @LennartPfeiler ,

Thank you @Shravan133  very much for the solution,  to help you understand the problem, I found a document to help you understand:

For your question, I found the relevant documents for you to solve your problem, in Power BI, mainly divided into two types of model prediction methods,

one is ETS, one is ARIMA.I in Power Bi desktop, I use line graphs, to achieve some prediction effect, I hope it will help you.

Here is the document I found for you, I hope it helps.

Describing the forecasting models in Power View | Microsoft Power BI Blog | Microsoft Power BI

Hope it helps!

Best regards,
Community Support Team_ Tom Shen

If this post helps then please consider Accept it as the solution to help the other members find it more quickly.

[LennartPfeiler]

Thank you!
But I can`t find the source where it says, that ARIMA is used. There is just a reference. 
Where did you get this information?

[Anonymous]

Hi @LennartPfeiler ，

This is the information I found myself about power bi's predictive model regarding ARIMA, I hope it helps you.

Time series Series with Power BI- Forecast with Arima-Part 12 - RADACAD

Hope it helps!

Best regards,
Community Support Team_ Tom Shen

If this post helps then please consider Accept it as the solution to help the other members find it more quickly.

[Shravan133 · Accepted solution]

Power BI's integrated forecasting feature in the analytics pane primarily uses exponential smoothing (ETS) for time series forecasting. More specifically, it employs an algorithm known as ETS (Error, Trend, Seasonal) decomposition.

Details on ETS Model:

1. Error (E): Measures the extent of random variations in the data.
2. Trend (T): Captures the underlying trend in the data.
3. Seasonal (S): Identifies and measures seasonal patterns within the data.

Two Models in Power BI Forecasting:

Power BI uses two different methods depending on the data characteristics:

1. ETS AAA (Additive Error, Additive Trend, Additive Seasonality): Suitable when both trend and seasonality patterns are additive.
2. ETS AAN (Additive Error, Additive Trend, No Seasonality): Used when seasonality is not detected or is negligible.

Documentation and Sources:

• Microsoft Official Documentation: Microsoft provides detailed documentation on Power BI’s forecasting capabilities and the underlying algorithms. You can find this in the Power BI Documentation on the official Microsoft website.
• Technical Whitepaper: For a more in-depth understanding, you can refer to the technical whitepaper on Power BI Advanced Analytics, which explains the forecasting techniques and models used in Power BI.

When Each Algorithm is Selected:

The selection between ETS AAA and ETS AAN is typically automated based on the characteristics of the data:

• If the data exhibits a clear seasonal pattern, the ETS AAA model is selected.
• If the data does not exhibit seasonality, the ETS AAN model is chosen.

Practical Steps to Optimize Forecasting in Power BI:

1. Preprocess Data: Ensure your time series data is clean and appropriately preprocessed.
2. Configure Forecasting Parameters:
   • Use the Confidence Interval to adjust the forecast's certainty range.
   • Adjust Seasonality manually if you have domain knowledge of the data.
3. Validate Forecast: Compare the forecast results with historical data to validate the accuracy of the model.

By understanding the underlying ETS models and their selection criteria, you can better optimize and interpret forecasts in Power BI.

[LennartPfeiler]

Thank you very much!
I´m getting on both links a 404 error. Could you maybe provide the links again? 
Thanks 🙂