Power BI - Statistical analysis (distribution analysis)

Power BI
1

Hi @EnterpriseDNA ,

I am was wondering if there is any available resource around statistical analysis in Power Bi.

What I am looking for is how to perform a distribution analysis in Power Bi.

Is there anything available on the platform? Maybe in the Power Bi Challenge we have something like this?

If not, that would be a great addition to the existing content here (I work as a Data Analyst and I am getting asked more and more often about statistical oriented analysis)

Thank u as always
Diego

1 Like
2

Hi @cassarodiego2
The only course within the Education portal that is

image
image437×690 91.5 KB

I know that EDNA experts is working on some more courses related to this to issues later, but I’m not sure when that will be done.

I hope this helps.
Keith

1 Like
3

Thanks @Keith ,

I have watched that course!!! Really well done!!

What I was hoping for was a more “Sam-way” structured course on Power Bi but more focused on Statistical Analysis w/o being statisticians. :slight_smile:

For instance, one of my clients asked me to study the distribution of some data, extract some descriptive statistics (mean, median, mode, standard deviation, skewness ect) and provide recommendations. I was doing this type of analysis in Excel which has the “data analysis” tool but my issue is with the size/volume of data to be analyzed.

That is why it would be awesome to get this type of course: statistics with Power Bi. That would be amazing.

Thank u for answering :slight_smile:

4

Hi @cassarodiego2

I’m sure that EDNA experts will likely have that already in the process. I guess patience is the key. I was listening/watching one of the podcast between Brian @BrianJ and Sam on youtube https://www.youtube.com/watch?v=t4GWCLoPUOk that Brian mention the Statistic for Data Analyst was being address.

I’m not sure what is all involved.

Maybe @BrianJ could expand on this.

@cassarodiego2 could you please mark my name “solved” under my name please. So i can get credit for part of the solution.

thanks
Keith

thanks

2 Likes
5

Hi @cassarodiego2

Keith is right, and in addition to that you can check also:

https://www.daxpatterns.com/statistical-patterns/

Best regards,
Maja

6

Hi @mspanic ,

Thanks for the link. I will take a look (this community rocks !!! )

@Keith : I am not sure if I did it correctly (marking your name is solved). Pls check and let me know.

Thanks
Diego

7

@cassarodiego2 yes thanks you did mark it properly

@mspanic thanks for the additional site. I have also marked it to check out :slight_smile:

thanks again

8

@cassarodiego2 ,

We haven’t done that course because as huge fans as we are of Power BI, it is just the wrong tool for statistical analyses. Take a look at this post I did last week - the type of Exploratory Data Analysis report that your client is looking for literally took me 35 seconds to produce in R with 2 simple lines of code. Reproducing the same thing in Power BI would take me hours or even days.

Here’s another example - MS MVP Greg Deckler implemented a Holt-Winters timeseries forecasting model in Power BI. I took him 3,168 lines of DAX and 140 tables. I developed the exact same model by calling the following six-line R script from within Power Query. It took me about 10 minutes…

library(sweep)
HW ← dataset
HWts ← ts(HW$Column, frequency=12, start=c(2017,1))
HW_MSHGQM ← HoltWinters(HWts, alpha=0.716, beta=0.029, gamma=0.993)
HW_MSHGQM.pred ← predict(HW_MSHGQM, 24, prediction.interval = FALSE)
df ← sw_tidy(HW_MSHGQM.pred)

The same results are achievable in Python if that’s your preferred route.

And it doesn’t require a lot of time or additional study - literally you could just take @gjmount (George Mount’s) intro R course for Power BI Users on our platform or @gmh1977 (Gaelim Holland’s) intro Python course for Power BI Users and just Google the rest (or run the code I provided in my LinkedIn post above).

If you really want to take the shortest route, Andrew Gard just released a YouTube video this week that teaches the basics of R in under 40 minutes:

I hope this is helpful to you. Please give a shout if you have any additional question.

Thanks!

  • Brian
1 Like
9

thanks @BrianJ for adding your comments

Keith

1 Like
10

Hey @BrianJ ,
That is all fantastic! Never been happier to have renewed my yearly subscription on EDNA.
What you said makes perfect sense. I am bit biased towards Power Bi but sometimes it is simply not the best tool.
There is one humble suggestion about a course / webinar that you guys could cover: how to run a proper data analysis from data exploration to the visualization part (unless there is something like this already).
The content on the EDNA is huge and sometimes a bit overwhelming: that is why a structured course that goes thru the main / key steps to approach every type of data analysis would be a great addition.
For instance: Step 1) data exploration (what type of data exploration? what type of descriptive stats to keep in mind and look at? How to interpret them? what tools are best suited for (R/Python/Excel); Step 2) Data Load and data cleansing, ect. You got the drill :slight_smile:
Thanks
Diego

1 Like
11

@cassarodiego2 ,

Thanks for the feedback - always wonderful to hear folks getting great value out of their memberships.

We don’t have anything exactly like what you mention above (though it’s a good idea I will discuss with Andrew after he finishes the third course on multiple regression), but I can point you to a few things that I think will be very helpful.

The first is an extended talk I gave at the Azerbaijan MyDataSummit last year entitled “Analysis Planning: The Oft Overloooked Pillar of Effective Power BI Reporting”

This session work through how to integrate Exploratory Data Analysis and Analysis Planning to provide a clear roadmap for analyzing a new dataset.

On his excellent Equitable Equations YouTube channel, Andrew Gard has a number of phenomenal “over the shoulder” data analysis videos, where you get to watch and listen to his thought process as he breaks down a different dataset each time. Here are a couple of my favorites:

Although he does his analyses in R, the general approaches are applicable in multiple platforms.

I hope this is helpful to you.

  • Brian
1 Like
12

I love this @BrianJ .

Yes Enterprise DNA helped to get promoted and stuff so I call myself a ‘very high engaged and satisfied customer’.

Your talk was very clear so thank you for posting it here for me. Very helpful indeed.

Diego

1 Like