Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 

Register now to learn Fabric in free live sessions led by the best Microsoft experts. From Apr 16 to May 9, in English and Spanish.

Reply
Anonymous
Not applicable

Forecasting in Power BI (The Theory)

Hi Everyone, 

 

I'm hoping that there might be someone out there that could point me to a few resources (Power BI Forecasting for Dummies?) or some kind of starting point so I can do a bit more research into the "How Power BI Forecasting" works side of things. 

 

Essentially, I need to explain (to management) how Power BI does it's forcasting. At the moment I've told them that it uses "soothsayer" technology as a delaying tactic but I think they'll eventually figure out that it's not a thing.

 

I'm not really looking for the: here's the full mathematical functions of how it works but even if it's the name of the mathematical theory that's being used, I'm happy with that too.

 

Thanks in advance.

1 ACCEPTED SOLUTION
Anonymous
Not applicable

I've found what I was looking for:

 

https://powerbi.microsoft.com/en-us/blog/describing-the-forecasting-models-in-power-view/

 

TL;DR

 

Two versions of exponential smoothing.

1) One for seasonal data (Power BI variant of Holt-Winters)

2) One for non-seasonal data (Simpler equation to model the time series)

 

Appropriate model used automatically based on an analysis of historical data.

View solution in original post

10 REPLIES 10
djjeck
New Member

Hello everyone! 

Please, someone, help me. I have next task:

it's necessary to analyze some data in excel format (look picture, this is actually just an example) and predict in Power BI Amount of People.

The seasonal prediction doesn't work for my case, as I have yearly data.

Can someone explain step by step how to make forecast using Non Seasonal method for yearly data sets?

 

Sample.png

Anonymous
Not applicable

Seasonality isn't necessarily a function of a calendar year. You can have seasonality within a single month so long as the data's changes are regular and predictable. How are you testing for your seasonality?

djjeck
New Member

Hello everyone! 

Please, someone, help me. I have next task:

it's necessary to analyze some data in excel format (look picture, this is actually just an example Link to the pic ) and predict in Power BI Amount of People.

The seasonal prediction doesn't work for my case, as I have yearly data.

Can someone explain step by step how to make forecast using Non Seasonal method for yearly data sets?

Thank you!

Anonymous
Not applicable

I've found what I was looking for:

 

https://powerbi.microsoft.com/en-us/blog/describing-the-forecasting-models-in-power-view/

 

TL;DR

 

Two versions of exponential smoothing.

1) One for seasonal data (Power BI variant of Holt-Winters)

2) One for non-seasonal data (Simpler equation to model the time series)

 

Appropriate model used automatically based on an analysis of historical data.

v-shex-msft
Community Support
Community Support

HI @Anonymous,

 

I'm also not familiar with forecast calculation logic, maybe you can refer to below link about dax forecast logic:

Introducing Forecasting Logic Into Your Power BI Models Using DAX

 

According to official blog, I know power bi forecast feature is based on 'built-in predictive forecasting models to automatically detect'. I don't think microsoft will direct share these built-in forecast calculation logic and mathematical functions.

 Power BI Desktop September Feature Summary

The forecasting feature utilizes built-in predictive forecasting models to automatically detect the step (monthly/weekly/annually) and seasonality in your data to provide forecasting results.

 

Regards,

Xiaoxin Sheng

Community Support Team _ Xiaoxin
If this post helps, please consider accept as solution to help other members find it more quickly.
Anonymous
Not applicable

Thanks @v-shex-msft - I thought that might have been the case and just wanted to check with the Power BI community and other subject matter experts.

 

It's unfortunate that it's a "black box" type situation because trying to convince stakeholders of predicted trends aren't going to have much credibility when they ask me about it and I respond with "because Power BI says so" 😛

Anonymous
Not applicable

I should also chime in and suggest that you should look into R.  Power BI is able to use the machine learning capabilities of R.  While i haven't used it, R stood out to me as being a perfect platform for forecasting.

Hey @Anonymous ,

 

Do you have a workbook you could could share?

I'm looking to do Holt-Winters with R to produce a caluclated column in PowerBI. Any help would be appreciated.

 

Thanks!

Anonymous
Not applicable

@chriswragge  as mentioned in my post, i've not personally used it.

Anonymous
Not applicable

Thanks @Anonymous, it was more for me to be able to tell my managers which technique was used (e.g. linear regression or time series or anything really) and also so I could get a basic background of what Power BI was doing to come up with the numbers.

 

The "cos PowerBI says so" reasoning doesn't really help me build credibility.

 

I'll look into R and see if there's a custom visualisation that's a little more open about what predictive technique it's using.

Helpful resources

Announcements
Microsoft Fabric Learn Together

Microsoft Fabric Learn Together

Covering the world! 9:00-10:30 AM Sydney, 4:00-5:30 PM CET (Paris/Berlin), 7:00-8:30 PM Mexico City

PBI_APRIL_CAROUSEL1

Power BI Monthly Update - April 2024

Check out the April 2024 Power BI update to learn about new features.

April Fabric Community Update

Fabric Community Update - April 2024

Find out what's new and trending in the Fabric Community.