Learning Python – Ep. 216

Python is showing up everywhere in the Microsoft data stack right now—Synapse notebooks, Azure functions, even scripts inside Power Query. So it’s a fair question for any Power BI professional: do I need to learn Python, and if so, when?

In Ep. 216, Mike, Tommy, and Seth use the Power BI + Jupyter notebook story as a jumping-off point to talk about the real role Python can play in analytics teams: an upstream workhorse for data shaping and automation, a flexible sandbox for exploration, and (sometimes) a distraction if it pulls you away from the fundamentals that make reports usable.

News & Announcements

• Add a vertical bar to your slicer — A quick UI pattern for improving slicer readability when you want clearer visual separation between items.
• Try a different Power BI theme in the Service — A handy workflow for experimenting with theme files on existing reports without rebuilding everything from scratch.
• Create Power BI reports in Jupyter notebooks — The announcement that sparked today’s conversation: notebooks as a first-class place to explore, transform, and then build reports.
• Submit an idea/topic for the podcast — Got a question you want the crew to tackle? Drop a topic suggestion here.
• Friends of PowerBI.tips: Argus PBI — Community/sponsor link from the episode notes.

Main Discussion

The core theme of the episode is role clarity. A report builder and a data modeler can be absolute pros in Power BI without writing a single line of Python—yet Python keeps showing up in job descriptions because more organizations are pushing data preparation upstream into notebooks, pipelines, and engineering-friendly tooling.

The team also compares the notebook development experience (cell-by-cell execution, fast iteration, strong editor tooling, Copilot/IntelliSense) to more traditional environments. The point isn’t that notebooks replace SQL or Power BI—it’s that notebooks are becoming a common interface for the work that happens before your semantic model.

Key takeaways from the episode:

• Don’t treat Python as a replacement for DAX or Power Query; think of it as an adjacent lever that expands what you can automate and how you can prepare data.
• Notebook workflows are great for incremental development: run one cell, validate results, document intent, then move forward—especially for exploration and repeatable prep steps.
• SQL still matters (a lot). Python complements SQL when you need richer libraries, file/API handling, or non-relational manipulation—not when you just need a clean SELECT.
• If you’re early in your Power BI journey, prioritize modeling, DAX, and Power Query first; Python is most valuable once you’re bumping into limits of copy/paste ETL and manual processes.
• The Microsoft ecosystem is putting Python in more places (notebooks, services, automation), which lowers the barrier to entry and makes “being able to read a notebook” a practical career skill.
• Start small: a couple scripts for repeatable tasks (file transforms, lightweight API pulls, helper utilities) will teach you more than trying to learn every framework at once.
• The closer your org gets to cloud engineering patterns (Synapse/Databricks-style pipelines), the more valuable Python becomes—because it’s often the glue between raw data and the curated model you build reports on.

Looking Forward

If you’re a Power BI pro, keep leveling up the fundamentals—then learn just enough Python to be dangerous, because notebooks are increasingly where the upstream analytics work is happening.

Episode Transcript

0:02 [Music] foreign good morning everyone welcome back to the explicit measures podcast with Tommy

0:33 the explicit measures podcast with Tommy Seth and Mike hello everybody hello Mike how are you today hello Michael Michael Mike Michael oh Michael yeah well you hate him Mike I’m sorry I keep changing my nickname it used to be Mike DeCarlo no it’s no it’s just Michael that one that one Rings yeah Mike DeCarlo just flows off the tongue there when we start our Italian translation of the podcast yeah which would require a first learn Italian so

1:03 would require a first learn Italian so laughs laughs I know a couple words like spaghetti so jumping in today a couple intros a couple things across the internet so first off I’ll I’ll point out the Microsoft build conference is coming up next week next week do exactly which days are the conference I have it blocked off my calendar like Wednesday to Thursday when Wednesday Thursdays 24th 25th I think April 25th that’s those are the days of

1:35 April 25th that’s those are the days of the conference I think so unless it stretches the 26th I couldn’t remember if it was Tuesday through Wednesday I’ll have to Google it again but the Microsoft build conference is coming up if you haven’t seen it you’ve probably seen a number of very high level in people in the party organization touting it and pointing to it and saying there’s some big things coming so coming so definitely worth your time so next week stay tuned jump into the build conference Listen to I think the keynote

2:05 conference Listen to I think the keynote and there’s a lot of comments from people on LinkedIn and or Twitter that are highlighting some sessions you definitely want to attend and see so very excited for build you are correct Mike it is Tuesday Wednesday Thursday 23rd through the 25th in Seattle there we go online is only 23rd 24th there you go the first two days so Tuesday Wednesday are the online events I really wish I was trying to get to go I was trying to help volunteer but no one really needed my help so they’re like oh we don’t need you you’re just power bi get out of here you’re just a

2:37 power bi get out of here you’re just a power bi guy get out get out that being said some power bi things two blogs we actually want to talk about and and Seth you found this one from Eric Syverson it’s Fenton yes Benson I’m sorry I said this is a fun one and it’s a super quick super quick blog on his part but it’s always interesting to me that when clients come with an idea or your product teams or some UI ux element where they want to see something in a report and in

3:09 want to see something in a report and in his case he he had a client that wanted to have a a small color bar next to slicers that would visually pop you over into a particular area and as he walks through he’s like well the first solution is just create a little box right that has the the color and now I have two objects on the page but for every slicer I now need to create this box and then alignment all that stuff and what a fun little solution he found in there is using the shading or the shadow

3:39 using the shading or the shadow functionality for the slicer itself so he describes how you set up the properties of that to just have a bar of whatever color you would want on on the left hand side you could do bottom top whatever I just thought it was it’s one of those like oh now that I know that that’s super cool to know I wouldn’t I wouldn’t have originally thought of it so it’s a it’s a fun little fun little blog good little tip and trick I liked it a lot and I also thought as well I think he’s just using a

4:10 well I think he’s just using a shape shape next to the object yeah he was and I think you can control a shape color based on some data points right you can drive a shape based on some effects yeah so I was thinking potentially so look at this block I thought wow this is great I love the idea that you could have a slicer that has a color next to it and I also thought about I I dread this but people do it there are reports where there are numbers of filters on the same page or a filter section of the of the report on the left

4:41 section of the of the report on the left hand side on the right hand side and I could definitely imagine like okay there are similar things that you’re filtering like this is all customer level filters but you may put the same color next to them so these are all the same color filter or what you I think you could also do is have the filter driven by a function that if it is in context of the filtering filtering or it has it’s it’s somehow been changed or something light it up you could light it up you could have a gray and then when you light it up when you change something it actually lights up it gives you a little bit of a

5:11 lights up it gives you a little bit of a visual indicator there so just looking at his his layout and things there there is the function box next to the shadow so you don’t even need the extra well that’s on the shadow I’m not sure yeah that’s what that’s on the shadow but I think the background color of the square that he’s using I think you could do that too that too so yeah correct you could change the you could have it all driven by a function techno color yeah changing reports yes it’s incredible with all the

5:41 it’s incredible with all the conditional formatting areas I don’t think I’ve touched all of them because it’s it’s the ability to do the conditional formatting is on almost not every property but basically anywhere there’s a color I I’ve obviously touched a bunch but there are so many situations or there’s so many opportunities to do something I I need to spend more time on each and every one just to see what that can do but but it also goes too with if you have a client need where you’re like oh I would never think of trying that out

6:11 I would never think of trying that out yes exactly and before again us Oldies in this whole power bi space previously there was a function for formatting but it was never highlighted until they really made this new toolbar right you had to put your cursor over where you were changing the color property to see if remember that remember when when the effects button was hidden until you put your cursor over it and then it would appear and I’m like that is like the worst designed everything I’ve ever seen so I’m glad they fixed that part because that was like impossible to find

6:41 that was like impossible to find meanwhile if you thought about the filter pane or or the formatting pain you’d have to go through every property and put your cursor over it like does this format does this format I don’t know it’s like it’s like a it’s like a hidden hide and seek around will it format so that was interesting so I’m checking it out right now with the the filtering on the visual if I can actually change the color of the background yeah it looks like you

7:11 of the background yeah it looks like you can change the the it looks like you can change the background color of a shape to have a function next to it so yeah you could do that you could take the shape and make it formatting for this next to the slicer which would be cool neat idea he also he also did another speaking of colors he also did a another recent blog right on themes yes very true so Eric you’re on you’re on fire you’re you’re pranking them out here so here so we we have been using we have

7:42 we we have been using we have produced a new theme generator or updated our current one to enable you to create themes and then visualize what the theme looks like on top of visuals I for my awareness right now we’re the only theme generator that allows you to do that but sometimes you want to see the theme on top of your existing report you don’t necessarily want to see it on just a generic visual right so our theme generator doesn’t let you embed your own report and then apply the properties on top of that so what you can do is you can actually take the

8:12 can do is you can actually take the theme file or the things that are extracted from the theme generator you can copy and paste them and Eric walks you through how to very quickly make a simple theme and then go into the playground so power bi’s playground is for their developers and you can render your own report from your tenant inside the power bi playground awesome and then in addition to that when you render your own report you have the ability of being able to apply via properties of the developer

8:42 apply via properties of the developer features you can actually apply your own report theme so you basically can copy the Json directly into the web browser UI editor and then you can immediately see changes to the theme file or or what your report would look like after it has been styled with a theme so Eric jumps in here he shows you a demo of demo of here’s how you would change this file here’s how you would stylize the file using a theme file and then you could real time update that theme file as needed so very good blog I thought those are very creative way of being able to

9:12 are very creative way of being able to have a live view of your your report and your theme file all at once man I’ve been showing people the the theme generator and just the ability to save and showing the visuals it’s basically blown the back of people’s heads off it’s such a cool little easy way to do it oh my gosh yeah well the save the tool too like you see my my updates on GitHub I have been sending quite a few feature updates just because I’m using it all the time

9:42 I’m using it all the time and this is so important for what we do and especially that before you had to be back in Json if you wanted to do this level of customization honestly I would say the Json the or your theme creator should be required for power bi developers and to be honest I find that so I make tools for where I spend a lot of time right if I if I spend a lot of time and that was one of my problems themes came out great I like the idea of them but I spent so much

10:14 the idea of them but I spent so much time trying to figure out what property was doing what where this thing was

10:17 was doing what where this thing was going how to stylize this thing like and and now there’s thousands I think thousands of properties because each visual has hundreds if not tens of hundreds of visuals now and there’s just so much stuff you can do with all the theming at this point point it’s almost overwhelming so you almost need some tool to help you navigate what the theme thing is doing that’s another challenge I think I’ve also faced with the theme portion is people have tried to keep up to date repos of all the properties and all the

10:47 repos of all the properties and all the settings I know but desktop is now changing every month with new properties new updates new settings and so to stay up and current with what current desktop is doing is doing it’s almost impossible you have to be updating your theme stuff every month so now this tool is much more up to date it probably has the most the most complete list of all the theme properties at this point and I’m gonna just double check I think we’re releasing some new features this week week on there as well if you go into the properties area on

11:17 if you go into the properties area on the theme generator you now have a search window so now you can search for UCS so that just that just showed up this week so now you can then go search for an individual property and then it will highlight where those properties live inside the whole menu because that was in some cases if you’re looking at I was in some cases if you’re looking at there’s it’s a scrollable list of mean there’s it’s a scrollable list of items right so it’s very very long so if you want to search for something that makes it much easier to find a title or a header or the different objects that you’re looking for in there as well so that’ll be very much more helpful more features coming

11:47 much more helpful more features coming along the way as well stay tuned we’ve got some other big updates planned all right with that any other openers that we have here before we get into today’s topic looking for an answer here either one I think anyone here Bueller Bueller did you do it yet Seth did I do what yeah it’s Tuesday

12:19 yeah it’s Tuesday it’s not it’s not Tuesday oh my gosh are you serious I’m on the wrong day welcome to Thursday wow what’s our agenda today man I am it’s been a busy week apparently for me wow I thought it was Tuesday I could have swore I was like wait for Seth to say this whole time the first 12 minutes of this thing 100 to say it maybe that is welcome to not Tuesday Tuesday maybe that’s my Thursday opener my goodness I would very much check your

12:50 goodness I would very much check your calendar today oh yeah yeah I should make sure that I’m not gonna miss meetings now because I’ve I’m sorely off date anyways oh boy goodness gracious but hey I appreciate you teaming that up for well I was just making sure we didn’t miss it and just make sure just making sure we didn’t miss it miss it sorry on to the main topic for today Tommy you came up with this one and this is something that you’ve been learning a little bit about or trying to dive in a bit more and I will say this is being in the data

13:20 I will say this is being in the data space I feel like has been like a constant learning exercise always like everything I’m doing everything we’re working on there’s always something new coming out that I’m trying to figure out how does this fit what does this make does this make sense should I invest my time on this new thing and so on the Microsoft blog they put out a a feature here is create power bi reports now in Jupiter notebooks which I find very interesting so I’m going to put the post here or the main topic

13:51 put the post here or the main topic today so this is from the the Microsoft blog and then talking more about how do you embed things so looking through the article here and thinking through things one is this concept of a notebook right so I’ve done a lot of work inside databricks that’s where I love building things or data architectures at this point and Jupiter notebooks are all a part of that system or the ecosystem and I’ll have to admit here so one thing I just want to discuss here is that there’s a new way

14:21 discuss here is that there’s a new way of computing information I think we’ve moved away a little bit from a single machine being able to compute your data and now we’re moving into this massive parallel processing ecosystem right so databricks is the system that does a lot of this really cool cool Notebook based stuff and so I I came from ssms and then I see Jupiter notebooks notebooks I really like developing in Jupiter notebooks a lot and with this goes with this General programming linkage called

14:52 this General programming linkage called python python so initial thoughts what do you guys think about the blog what do you guys think about this concept of like notebooks and then you concept of like notebooks and then joining that with our data like so know joining that with our data like so notebooks traditionally came from like the data science Realm and now here we are joining the bi space into that notebook Arena what are your thoughts there’s a bit of a history here too because I think you’re referring to a lot of that from the data engineering side totally and if you really take I don’t go the complete history but yeah

15:23 don’t go the complete history but yeah back when python was made actually their name actually because of Mighty python which I still love to this no way it’s really from I did not know that actually that’s the actual the word in the name is because of Monty Python I thought it was because of a snake well you can take a look I’m pretty sure I gotta look this up right now because I might just drop everything from from now on just to learn this language that’s awesome yeah but when I started the data analysis realm I was like okay what do I need to

15:53 realm I was like okay what do I need to learn and it was either r or python but it’s like okay this is much more data science this is much more data engineering I was doing more business analysis and obviously power bi came out there wasn’t too much Synergy or overlap at the time but if you keep looking at the story of the business now the business analyst business intelligence there continues to be more overlap or at least a little more relationship between what python can do with all the packages and what you can

16:25 with all the packages and what you can do in power bi or what those skills are obviously there’s data engineering which is probably pure python but I think there’s more for even just the business intelligence analysts where okay maybe this makes a lot more sense when I saw this article came out with the Jupiter notebooks I’m like okay man and to your point like I’m constantly looking like five years down the road two years down the road what do I need yes yes and to me this feels like to me looking at Power bi inside a Jupiter notebook

16:55 at Power bi inside a Jupiter notebook and the versatility that you can get around with notebooks it feels like a skill that yeah this could be something like General programming language doing common data manipulations whether you’re a data scientist a data engineer or even not a bi developer right right it seems to me so Universal across all those spaces and frankly Python’s pretty easy to learn I started learning Visual Basic it’s okay but I was able to record so I started with Excel I started recording macros in Excel seeing what it what it would Auto generate and I feel

17:25 what it would Auto generate and I feel like most of what I’ve learned from languages either SQL or how I started learning anyways I’ve gotten a little bit more sophisticated than recording something right access databases is how I started learning how to write SQL I would grab a couple tables I would pull a couple together and I would then put the little code view oh look look at what it did oh interesting this is how it did a left join oh here okay that’s interesting so I was actually using like the UI to help me generate a little bit of code that then I was able to dive in further and

17:55 then I was able to dive in further and go deeper with that I’d be curious Seth from your perspective so the reason I’m I’m bringing up this whole ssms experience developer-wise and Jupiter notebooks development wise I feel like those are the two ssms felt older school to me where I was writing a SQL script and it’s a a single like text file almost where I could highlight a portion of text and then run it whereas in notebooks or Jupiter notebooks you have the ability to be able to run a cell of information so

18:27 able to run a cell of information so it’s almost like a it’s almost like an evolution of like a sequence of thoughts right you can start thought one thought two thought three and I remember when I was writing some more SQL statements in ssms I would do that I would build you ssms I would do that I would build a couple temporary tables at the know a couple temporary tables at the beginning maybe in the script I would set my variables at the beginning of the script and then I would work my way down the script to see what it was producing and I would maybe do like a couple like I I’d had to highlight the middle section and then run it to see what the what the output was but it it feels the same but it better

18:59 but it it feels the same but it better in in a in in a how my mind thinks and how I like to write and build and develop code yeah I write and build and develop code yeah obviously the interfaces are are mean obviously the interfaces are are more more newer right so they’re they’re evolving with the code sets especially as you can with the jupyter notebook switch context of code right that’s true that’s very true like the first several cells I can do some transformation and I can I can say okay now I want to like interact with that using SQL so I think

19:29 interact with that using SQL so I think the capabilities are are a lot better in in the uis and and how we can write code to do what we want it to so in context like what is your original question again like so my question yeah sorry I guess my question was really more around like looking at how you develop an ssms and now comparing that to how you develop in notebooks my personal preference is whether I’m writing a SQL code block or just python code blocks I really like

20:01 just python code blocks I really like the idea of running like the cell based notebooks that you can run with a Jupiter notebook right the the cell based one step at a time so I prefer the now Jupiter experience over developing on ssms for anything code wise yeah I don’t I don’t know like there there are components where depending on how long the code gets within a particular cell that I’ve like in SMS like just executing certain parts of code blocks is within that context is

20:31 of code blocks is within that context is easy right because you just highlight

20:33 easy right because you just highlight and execute that that portion sure whereas whereas Jupiter I don’t think databricks does that well or Jupiter notebooks yeah it’s like you get you get all the cell or none of the cell basically right so but then I just build more cells and usually in general yeah I I would say I’m I’m biased right because I’ve been using Jupiter notebooks a lot more in the last four years yeah yeah as opposed to just straight SMS but across the board I I wouldn’t I wouldn’t say

21:04 the board I I wouldn’t I wouldn’t say like I would say I would temper this conversation with like we are we are all playing in Realms of big data right where large volumes of information these tool sets are incredible and allow us to do quite a bit but I don’t I don’t think that’s the majority of companies yet right so like I I think what will be interesting is comparing and contrasting like the need for learning python or all of this within Jupiter notebooks with the contrast of SQL still is the king of of

21:35 contrast of SQL still is the king of of database systems and like there are a lot more people writing SQL in servers and and things that aren’t touching the capacities that we need to or the size or whatever need to that’s a that’s an interesting concept as well there I have to think about your comment there a bit more so I I think let me let me let me rephrase if if our general hypothesis within this conversation is hey python is a language that like should be a stepping stone for

22:06 that like should be a stepping stone for a power bi Pro I guess I would say two things things what is your definition of a power bi Pro because this definitely moves into bi developer realm okay you are a hardcore in the Enterprise Tech space this is not the business user is it is that agreed or argue that and I think that’s where my the Crux of my question is is a lot of the conversation you guys are having the dialogue is for the data engineering someone who already has that set up I think I’m making the argument or the CR

22:37 think I’m making the argument or the CR the origin of my question is for a power bi developer Dax m is now python going to be part of that or Jupiter notebook is going to be part of that not so much like the business user but I think for someone from my Persona who’s been focused so much solely on either SQL but really focused on the front end more it seemed me more and more that Python and jupyter notebooks are seeping into that request that’s what you’re saying but it depends it depends on the ecosystem that you’re

23:07 it depends on the ecosystem that you’re playing in right like how many how many companies in in business business folks you’re saying business folks well I’m saying the power bi developers like power bi Pro I wanna I wanna add another flavor here to Tommy’s comment though I I would I would want to break your power bi Pro into a report developer professional and a data modeler professional and I would I would argue the data modeler and the data engineer that role is getting very blurry to me at this point right

23:40 blurry to me at this point right I I feel like there’s a lot they may not be the same person but in some organizations they may be the same person and so I feel like there’s a very gray line between I’m a data engineer and I’m a power bi data model because a lot of the skills I think are mainly transferable across it so I just want to bring up the so I would I would I’m trying to bring up the point of I think in a power bi Pro or power by professional there’s that there’s there’s two rules so I didn’t mean to represent I’m sorry but like I think if

24:10 represent I’m sorry but like I think if we talk about those two roles I think it helps me understand like or where I would argue where you need to learn sorry what do you call it one of this power bi Pro and the other is what so in a PowerPoint professional there is a report Builder and there’s a data modeler and I think if I distinguish between those two roles it could mean it could be in the same one in the same person but if when I’m talking about probably a professional I’m talking about I think what the when we’re talking about python we’re talking about data engineering more exercises I think we’re talking more about that power bi report or the power bi data

24:41 power bi report or the power bi data modeler right and I want to make the distinction that this person probably wasn’t doing engineering or working in like synapse before they’re purely like I guess I guess think here here’s where I get stuck okay is is a could a super user in a business unit be a power bi Pro yes 100 they could be modeling things and they could be like and this is where that falls apart for me right because on power bi Pro can absolutely be a pro of connecting to multiple

25:12 be a pro of connecting to multiple different data sources using the tool to Empower query to extract like transform model data and build reports and and be a pro at Power bi does that mean they have the skill sets around larger data sets or using python or Jupiter no they’re not going to touch that stuff but they can be very very they can be a pro like they can know power bi backwards and forwards like we’re talking about data engineering aspects of Big Data of where you’re interacting

25:43 of Big Data of where you’re interacting with cloud services yeah right oh pick your tool right cloud services python in anything like you have to be in the cloud to do that yes and no right so you can run python locally you can run notebooks locally but that’s not going to be something that’s going to be part of your normal Power bi pipeline right that’s so you Power bi pipeline right that’s so there is there is this level of know there is there is this level of notebooks today are in synapse and above right so this this is an exercise in

26:14 right so this this is an exercise in data engineering that occurs just before you get to the data modeling space right so if you if you’re bringing you’re doing that data engineering work outside of the power bi ecosystem right so you of the power bi ecosystem right so if I’m interacting with python or know if I’m interacting with python or or building Jupiter notebooks whatever that is you’re either using synapse or data bricks that’s what that’s your options so in lieu of that right so I think Tommy’s pointing out here is is this a skill that even if you’re in the business unit thinking about these things like is this

26:45 thinking about these things like is this something on the horizon that you feel like that person should know and I would argue argue it’s I think it’s a bit more dependent on what you have access to in that business unit right well let me let me ask let me ask a clarifying question because I’m getting challenged in in like maybe in the chat right why do Jupiter and python immediately make y’all talk about Big Data okay let’s not talk about it I would agree with that I don’t think it has to be in a big date of conversation talk about cloud services though right isn’t that part of it or are there solutions that can be built in Python for ETL in Frameworks

27:15 built in Python for ETL in Frameworks on-prem I don’t know this question well I think part of my argument too is maybe not where we’re at right now oh I’ll answer it tell me what’s going somewhere else so well no I okay go ahead so I I think I think the Jupiter notebook space turns a lot more towards a conversation around the type of compute engine that you’re using to manipulate the data right so you’re not using jupyter notebooks on top of SQL right because there’s there’s no really concept of like a Jupiter notebook and there maybe there’s some in handling but I think it’s more of a

27:46 handling but I think it’s more of a spark based thing right so it comes it comes out of like the Hadoop area it comes out of spark but now if I’m talking spark and Hadoop am I not talking big day why would I invest in that if I don’t have large data volume those I think Hadoop would start with more of that big data conversation but I think now that we talk with spark I think spark is less motivated by large amounts of data I think you can run it efficiently or even at a again I think it goes to cost right at a certain point you could just spin up a SQL server and throw some

28:17 just spin up a SQL server and throw some data at it and then the cost to run that is minimal until you hit large volumes of data as soon as you hit larger volumes of data turning on a SQL Server becomes more cost prohibitive if you’re having like you prohibitive if you’re having like terabytes or hundreds of terabytes know terabytes or hundreds of terabytes of data in your in your ecosystem right so you would not be practical to put a petabyte of data in a SQL Server you probably could do it but you probably wouldn’t want to I feel like there’s a bit that we need to take a step back with because I think

28:47 to take a step back with because I think we’re all talking about the here now or the last two years what technology is available and what roles there are if we continue to see how at least technology continues to evolve and look at Power bi even outside of python the tools available for us the data modeling and the web more things that are accessible I would make the prediction that with what’s going on with Jupiter notebooks with what the role of the power bi Pro is go back five years I didn’t have to know half the stuff I know now I I have

29:18 know half the stuff I know now I I have to know the data monthly web and the publishing and probably governance too to me the technology is only going to a place where that’s going to be part of my requirement to at least be able to do not necessarily the whole full stack of python or Jupiter notebooks but I think if you were to look at a resume or a job description three years from now and if you want to get into Power bi I feel like Python’s going to be on that job description

29:50 that’s a good question and I almost would make the argument more than SQL well so today you can already use Python inside power bi by making like some data Transformations inside M power and inside power query you can you can write your own block of code or either I think you can do r or python you can add little python or our scripts directly inside your M code which I I have found so for me personally right writing Python scripts inside the editor of M one we don’t have great M intellisense anyways and then to add python to that

30:20 anyways and then to add python to that you’re not even getting great python Intel sense inside sure the the M experience more tools accessible vs code Jupiter notebooks oh my goodness with data Wrangler now there’s so many ways there’s so many things that are coming out like that for me for being a python newbie my growth path because it’s such a popular language has been exponential since I’ve really understood this because of the intellisense and the ability and co-pilot where it’s like okay I have I

30:50 co-pilot where it’s like okay I have I feel like I have no excuse not to know it and how it correlates I’ve been well again again the conversation of is pi so let’s let’s take a little bit more of a step back right so I’m going to take a little bit of a higher view here right okay we’re talking about power bi Pro or property professionals right now at this point but if I step back and look at other tools that Microsoft provides to us like for example power automate you can do automation scripts on things and there are some things that power automate can’t do on a desktop but you can solve with Python scripts that do very simple

31:20 with Python scripts that do very simple manipulations to images or other things there that you want to do so like if you look at like where Microsoft is applying where you can apply python right they’ve given you synapse they’re giving you notebooks they’ve added like so where do I see like this is where you have to say I don’t I don’t want to listen to what Microsoft says I I look at what they’ve been developing like where are they putting their time and their effort right they’ve been incorporating more Tools around that power bi ecosystem so synapse has it power bi has it

31:52 synapse has it power bi has it Power automate has a little bit of python in it so I think in General right if I’m that business user and I want to learn something that’s valuable Python’s very well documented there’s tons of examples and frankly if you go to copilot and you need a little python script copilot or get or Bing or Bing chat whatever if you just type your question there how do I XYZ in Python it’s pretty dang good to write code for you to help you learn and it even it even does a good job explaining it it’s like this code’s doing this and in this

32:22 like this code’s doing this and in this line doing like oh my gosh like so for people to be able to start joining like that python crew I think the barriers to join is becoming less and less right and where I see Microsoft going is they seem like they’re putting it in more and more places to make it easier to use and yeah and I completely read it and not even from a data engineering point of view Python’s always been an incredible tool just for the data scientist or for business intelligence yes but too often because power bi didn’t play not well with it but there

32:53 didn’t play not well with it but there wasn’t that overlap and again the accessibility well those barriers are coming down or they they seem to at least and someone else is mentioning chat here right not to mention also Azure data Factory also lets you use python as well so so I would say directly or through Azure functions I I believe it’s I don’t know I have to look it down so I think you can use it I’m not aware of I’m not aware of a direct way to do it in ADF like outside of

33:23 way to do it in ADF like outside of using Azure functions which maybe has to be an Azure function based thing but but regardless though I I feel like this is a skill that one it’s valuable to learn it it’s being incorporated into many different parts of the tools around data and data intelligence and I think a natural progression would be okay I’m in the business I’m learning M I’m learning visualization building that makes sense to me I think if you are dabbling in

33:53 to me I think if you are dabbling in power automate and again this is another place where I think there’s potentially synapse is like the first product that I really feel like there should be some it developers in there and there may also be some business users in there right so we’ve always had this conversation on the podcast around hey I just need access to data I want to write a script against something and produce a table whatever that may be and I feel like synapse is one of these first tools where I can give regulated access to a team of people so in inside Azure so this is a the synapse

34:25 in inside Azure so this is a the synapse environment is really like the all right I’m gonna I’m gonna draw a line here because I think Seth you’re giving me some frowns here so maybe you’re just thinking about it but as I think about what is the business user’s realm of tooling and what is the it realm of tooling I think I draw the line around Azure there’s not as many business units I think dabbling in Azure now granted there are some that are doing it there’s some that are not but I think the majority of the it organization is focusing themselves inside Azure and that’s what they’re

34:56 inside Azure and that’s what they’re comfortable with spinning up servers working on cloud infrastructure the business isn’t as comfortable and when I was in the business we were like deathly scared of azure because it was like we don’t know how much it’s going to cost we don’t know like all these things we don’t know what we’re doing but in reality you could spin up a SQL Server let it run for a month and have minimal cost to it so you’re not really absorbing too much cost around that so from that perspective when you start going down that path think okay it’s not as scary as I thought it was when did

35:26 as scary as I thought it was when did you start getting into it then you start having this conversation of okay maybe synapse is that tool that heart starts bridging taking those business users giving them a bit more capability with SQL tables database structures that are in a SQL serverless environment that does start exposing notebooks to you and so I think allowing the business to start playing in that again I would still feel like I argue synapse still feels like a power bi-like tool all right it’s not like I’m

35:57 bi-like tool all right it’s not like I’m connecting to a server via ssms most of the activities you need to produce inside synapse are in the browser so I just feel like that’s that seems to me the way Microsoft’s going they’re not developing a ton more with ssms at this point they’re not developing a ton more with store procedures right they’re using other mechanisms to trigger and execute on top of data flows and I think the main reason why Microsoft keeps storages around is because there’s been store procedures for eons and they have

36:27 store procedures for eons and they have to keep them around for support for legacy but I don’t see them building like wow we built 10 new connectors inside or 10 new data engineering tools inside store procedures anymore or something along those lines I feel like that is not as much of a development cycle as what we’re seeing now a store procedure is nothing I I don’t quite understand that comment because a store procedure is nothing other than saying like I developed a Jupiter notebook to do a thing I can do anything with the store procedure I can have it return data I can do a bunch of ETL within it I can

36:58 can do a bunch of ETL within it I can transform things I can modify delete update I can do like that’s the mechanism by which I I execute a manipulation or an extraction of data within a SQL Server say saying like oh they’re not developing like it’s the framework are there are they are they increasing the capabilities within those tools to like read out of Json of executing python or like doing cross-functional things across like with other tooling yeah

37:28 across like with other tooling yeah absolutely like that’s the play and it has been for a long time because extending that tool from what what it has has served in a very for a very long time yeah right I guess we’re we’re I’m still hung up in this conversation is I see that we haven’t clearly articulated at least in my mind we’re talking about a wide swath of it when we say power bi there is a wide swath of users and if if we’ve talked about multiple different times in this show in in a priority

38:00 times in this show in in a priority Matrix of things you should learn when your job is developing reporting within an organization stacking python against Dax visualization M certain things is where like people not accustomed to listening to the show or can be like Oh I’m this business user I should go Learn Python and I don’t agree with that I don’t think to me this is part of one of those front and forward things that like falls into the or learning power bi

38:31 into the or learning power bi python should be like the next thing you learn and it’s important for me that we articulate like where that line is because I don’t like this is a data engineering part of a connection right connections connecting to data a transformation language right like that that’s a component of certain ways in which organizations have manipulated and done ETL which is a developer focused task typically correct yes so when we say power bi Pro like the delineation is we’re talking

39:03 like the delineation is we’re talking about development a developer level task so if that’s your interest area as a front-end business user who learn is learning power bi that you want to move towards the developer role would this be applicable as far as a language to learn if you didn’t know any would you yes would you prioritize that over SQL so this is a good question and this is where like this over SQL but I definitely think it’s high it’s High Mark like so I would for sure say SQL is

39:34 Mark like so I would for sure say SQL is probably your number one language if you’re if you’re doing things in data SQL it’s everywhere and it’s right and the reason because that allows you as a from the business pro from the power bi Pro perspective yes that allows you to connect to many more sources of information or like be it a sequencer beat Oracle be it my sequel be it like even data bricks right I that language traverses all of these different tool sets that I would be interacting with in many different ways

40:04 interacting with in many different ways I would change your phrasing just slightly and then I would agree with you you said the learning SQL gives you access to many more sources I’m not necessarily agree with that statement I would agree with using SQL gives you many more options to use a different compute right so I think the main the main difference for me here is am I using a SQL server for computing which runs SQL verbatim or am I using some other compute engine and it seems like the two that are the ones that are most interesting to me right now are

40:35 interesting to me right now are spark and SQL servers those are those are the two main engines that I interact with there’s not many other engines that I from a Computing standpoint that I integrate with that are as popular as they are sure I I think from my point it’s important to distinguish when I first started going into the bi space it had nothing to do with data engineering and it was very rapidly seven years ago it was wrapped with data science and

41:05 ago it was wrapped with data science and it was anything if you looked up business intelligence it was data science it’s like what do you need to learn learn yeah so honestly the data engineering side of python is relatively new to me that’s it came after the fact that if you want to be in the data space you need to know python to do analysis and express and to do a deep analysis had nothing to do with engineering right so there’s a whole other realm of python too that’s outside of spark that’s correct and outside of databricks where you may not even have to go into that

41:35 you may not even have to go into that space at all that’s very true and I I think for me that’s what I’m arguing yeah my Seth your original point if I’m the bi Pro or developer do I need to have to now go to the engineering side probably not unless you all want to become an engineer but for that and now for the bi Pro or the person who’s doing the analysis where there’s I think at least more of the blend with the data science too and again it’s only because technology is becoming more accessible and I can learn it much more quickly

42:06 and I can learn it much more quickly with solo pilot and be it so so what I would say is we’re we’re talking about power bi Developers sure because in my in my world the develop a power bi developer can do the whole Spectrum Enterprise and Enterprise you do ETL work you can do transformations in whatever system you’re working on like you you manipulate Data before you actually get to a modeling layer wherever you do that and you have those skills and capabilities to do that across the board

42:37 capabilities to do that across the board and organizations that are large enough do you have separate teams for this absolutely data Engineers data modelers visualization but in many places or those people can Traverse that ecosystem very very well right and if you’re only a SQL person right would I say python is one of those languages that you should add to your Lexicon I 100 percent right because even in in areas where I agree Scala which is a very smaller

43:07 I agree Scala which is a very smaller subset language much more like on a direct function language part though have like the experts that I’ve owned every one of them is if python you can pick up scholar right if you you can pick up scholar right if like and what this does is provides know like and what this does is provides a different framework of how you interact with ETL different Library like a whole host of features and functionality that just aren’t there from a query yes language perspective that I think make it extremely valuable plus it’s one of the like most popular programming languages out there and it’s

43:37 programming languages out there and it’s growing up python right I’m not saying that it’s not important I’m making a distinction of where it’s important sure and I think for me like I would get frustrated if I’m a power bi Pro and now like oh you got to know Jupiter notebooks to do more analysis it’s like oh I already have to be an expert in Dax and power query and data modeling and SQL and it’s now python going to be part of that stack I don’t know but it’s to me it seems like

44:07 me it seems like it would make sense and but it’s frustrating there is some there is a distinction I want to make that based on conversation that’s happening in the chat when I say learn SQL I am not saying learn the SQL engine there is a huge distinction between DBA and learnset EBA yeah or understanding the ecosystem of the entirety of SQL servers and no I’m not saying that like from a bi professional you need to learn how to query effectively and efficiently and above and beyond that like don’t

44:39 and above and beyond that like don’t waste your time like you’re not you’re you maybe some Performance Tuning stuff but like long gone are the days where I’m gonna say you need to learn the architecture of how SQL servers work and Views and like and indexes and how we create things like in general fine but you could waste an entire like a lot of time in delving deep deep what I’m talking about is this like SQL querying is your sweet spot like if how to manipulate a query and generate your

45:10 to manipulate a query and generate your output in an efficient way you’re done move on move on to the next language that’s going to help you like level up your skills and make you much more comfortable and I think this was the point that was being made by Mike and yutami was much more comfortable in these new ecosystems if you get the opportunity to interact with them which is Services especially in the Microsoft yeah right correct because 100 a ton of this stuff is you can Traverse a lot of this ecosystem with python

45:41 a lot of this ecosystem with python and it’s easier and I think all the books on data Explorer are data storytelling data analysis starts with that data exploration side and power bi is maybe not the best tool for the exploration seeing the outliers seeing I agree with that the different areas there it’s it’s you gotta have to do some work to get to a point where you’re like okay I can actually get to a place where I understand right what’s occurring in the data but data profiling tools are have been a little bit more common inside the the python space yes exactly exactly so and I think that’s part of what I’m

46:13 so and I think that’s part of what I’m saying too but yeah even when you look at the blog article so I’m going bringing it back here right so we’ve been talking about like python a skill set right jumping into a Jupiter notebook and then being able to quickly say okay from this data frame build me an auto-generated report that’s what they’re trying to make you do they’re trying to say here’s some data profiling on this data frame or whatever you’ve produced inside a report and you can continue to manipulate and pick different data points and then add them to the visuals on the on the report canvas there for power bi but in my mind

46:44 canvas there for power bi but in my mind I’m looking at this going like they’re taking the visualization side of power bi and applying it to data categorization some data investigation they’re trying to make it a bit more automated because in order for me to build four or five and again the example they showed there’s like six visuals on a page it takes a bit of time to figure out which Dimensions which measures which numbers are the right ones to fit together to show you actually some information about what’s occurring in your data set so I I’ll go ahead are you say something no I was gonna say I would agree

47:15 no I was gonna say I would agree completely because part of any data analysis is what do we need to filter out what do we need to group by what are the outliers right and like I think that’s always been more difficult in just power bi visualizations like yeah there’s some like I would always use like the box plot and the histogram to try to do that but it’s still a harder thing where literally like python packages have been made for that yeah I would agree with that and another another consideration here is I

47:45 another consideration here is I think what we’re also seeing here is we’re seeing a Divergence so to your point Seth earlier around what size organization are you in and what level do you have access to be able to do the data engineering right if you’re in a bigger organization and there is a department a team that is already owning and managing the data you probably are not going to need as much skill around the python space because there’s just not as much opportunity to get into to your point the cloud services like a synapse a databricks right where you’re going to be able to leverage that and make data engineering pipelines

48:15 make data engineering pipelines on top of that with python the the flip side of this is and I think this is where I feel like a majority a larger majority maybe I don’t know there’s like the medium to small side of the business or there’s the departmental space of this where we don’t have support from a broader it organization and there is a call it that the power bi data modeler and they’re not just level 100 anymore they’re level 300 400 right they’re very comfortable writing them they’re very comfortable doing a lot of things in there and maybe they’ve gotten

48:46 things in there and maybe they’ve gotten to a point where that individual will eventually or will work on adding other tooling like Azure data Factory or synapse they’re going to start venturing into a harder not a harder a more data engineering like role right so that data modeler expands their area of influence and they’re going a bit deeper and they’re saying okay the power bi data models and the power query I have today is not performant enough everything I hear says I should start

49:16 everything I hear says I should start doing those Transformations Upstream I’m trying to use roach’s Maxim right transform as far upstream and downstream is necessary so I think there becomes a point when you’re you’re at that the model builder level where you need to get more control around what those data queries are and where I have found that’s been very true in my career I started very much a very heavy power query lots of lines lots of Transformations and as I get into organizations where I am the sole data engineer did a modeler and Report

49:46 engineer did a modeler and Report Builder and helping them get their feet on the ground around power bi I have more control of things further upstream and so I’m helping them turn on a synapse I’m helping them build stuff up Upstream that now my power query becomes incredibly simple so my power query queries look like three or four steps now steps now and all of that data engineering work is being done Upstream in an engine that actually works a bit more efficiently for recursive transforms right

50:16 for recursive transforms right something I have to look up or weird joins or whatever that stuff is I can then be a bit more focused on that Upstream data transformation and preparing more data before I get to power bi and again to your point Seth right right purely a data engineering role that that was it’s definitely more in that space but what I see is again the trend I feel like is that data engineering role is coming closer and closer and closer to that data modeler and we may not be there today they may be today very distinct or two separate roles but I

50:47 distinct or two separate roles but I could very much see in the next year or two that that role becomes even closer together and we’re going to even see more connection between that data engineer role and the data modeler role they’re almost going to be synonymous right and I will make this point too right from from Trent from someone who reads a lot of resumes and works in a cloud cloud environment I I would say if you are a a like a a full stack developer of on-prem where

51:19 full stack developer of on-prem where SQL is was the thing

51:22 SQL is was the thing python is absolutely something that you should add to your your resume or your because that that puts you like what I’m trying to say is many many many resumes of folks that have been in this world for a very long time think that the three pages of how they managed hardware and infrastructure of SQL is impressive and it’s not because I don’t care about it yeah we don’t have to

51:52 care about it yeah we don’t have to support that in the cloud correct yes right so what are the languages that let you navigate and modify and manipulate data and I think this is to me the most interesting aspect of how things are evolving in this ecosystem it’s how efficient you can be with manipulating data not without the cons I’m not gonna say without constraints of like understanding closer usage and what you’re actually doing in ecosystems because there’s a cost however like less time needs to be spent on maintenance of

52:22 time needs to be spent on maintenance of things or setting up the structures yourself that’s a really these Frameworks are much more automated in a fashion of just handling that for you so if if there is a level up for data engineering years of Legacy not legacy platforms but platforms that are typically just more SQL based python would absolutely be one of those things that sets you apart and the last thing I’ll say too is even from from my point of view is just going with the pro or the the data analysis

52:52 with the pro or the the data analysis business intelligence the barrier to learn is learn is so low with the tools available oh my gosh that’s right any AI extension or any visual code extension that has to do with AI for copilot what does it always say start first works with python python is such a popular language github’s co-pilot chat chat GPT the the ability to translate I can translate t equal to python in in copilot and for

53:23 t equal to python in in copilot and for me like it’s been this exponential learning curve compared to learning other languages because it’s so popular and because it already integrates with so many other things that I do so like it’s almost I don’t say it’s almost doesn’t make sense not to but if it is going to be at least somewhat of what I do maybe not all of what I do even from the discovery and exploration and data science it it’s very hard to make the argument not to at least know the basics I’m Florida right now that this conversation went so fast in my opinion

53:54 conversation went so fast in my opinion I feel like I’m like 30 minutes in I’m looking at the clock going oh my gosh we’re almost at the end here so I guess I’ll probably have to give us a final thought here and we’ll we’ll hop on over to chatgpt and see if chat GPS would recommend and recommend python by itself so itself so Seth I I think I think what you said was incredibly profound and for people who didn’t hear what Seth said I’m gonna have to reiterate I’m gonna I’m gonna read let’s say it in a way that makes sense to me just just because I’m understanding what you said as a hiring manager

54:24 as a hiring manager someone who’s looking for skills to bring on board if you’re looking for people to help you build cloud-based services the the experience you had around managing servers and on-prem things is becoming less important in your eyes in order to be able to manage these new cloud-based services and I think there is definitely a trend in the market I don’t care who you are is to businesses are trying to get out of the business of managing internal

54:55 business of managing internal infrastructure on things period it the cloud provides you more Innovation you have more opportunities to optimize and tweak your pipelines and disclose Market yep and and so there’s there’s a there’s a very competitive move here to move organizations off of on-prem and into cloud and also I think what you’re also going to find here too the services that companies are purchasing companies don’t want to purchase a full sap server on-prem anymore they want to go buy the sap server as a service and then what is

55:25 server as a service and then what is happening is the organizations themselves the operational systems that run their day-to-day business are all moving to the cloud anyways there’s higher availability it’s less upfront costs it’s more amortized on the monthly basis and you it’s more of a pay to play method right if I use it more I pay more right if I sell more things I pay more which makes sense that scales with how well you sell stuff and so to your point here Seth like I think the trend here is again taking really high step back we’re going to Cloud

55:55 step back we’re going to Cloud in that cloud space if you are playing in that Arena having something on your resume that talks specifically around notebooks Jupiter Python and how that integrates with your business intelligence stack I think is very relevant from that per standpoint and I would also Echo the same thing right and again you’re Seth you’re hiring for people that are both front and front end and back end to some degree right you’re saying like I I use data bricks and I use power bi right that is now the stack by which you operate in and yes there’s a whole bunch of that that’s data

56:25 a whole bunch of that that’s data engineering but you’re looking for people who can span I I want that person that can span the data engineering space and can span the modeling area that’s the skill set I think we’re going to be looking for yeah ideally right but yeah even even in the climate that we are where the vast majority of people don’t have those skill sets that’s where it’s really important to be very very good at the transferable skill set yes right yeah so either you come in as as an amazing Dax modeler or power bi

56:55 as as an amazing Dax modeler or power bi visualization expert or the things that are so comfortable to you because you have to learn all these things you don’t know so I need to utilize you in the areas that you do know really well yes so whether it’s Azure data Factory synapse even even SAS versions of SQL right like all of that plays completely differently than on-prem Solutions that’s good that’s good anyways so that was the thing that really stood out to me it’s it’s it’s the cloud that is changing what we need

57:26 the cloud that is changing what we need to know and learn all right so quickly over to chat GPT as we wrap up here today so the question to Chachi P today is should I learn python for power bi development that’s the question I asked so the art the the question actually gives me like five good bullet points here learning python can be beneficial for power bi development python is a versatile programming language that’s widely used in data analytics data manipulation and data visualization tasks and then it lists out a couple key areas of where python does well right it does well in data manipulation it does well in

57:57 data manipulation it does well in advanced analytics for example machine learning you can use things like tensorflow Keras Pi torch like all these to our point it’s it’s a gateway to where you’re going to do machine learning it allows you to build custom visualizations there’s libraries that you can use that are built in like matplotlib or Seaborn these are extensive data Frameworks that let you visualize things in in Python now I would also Echo you could use those libraries inside power bi by making a python based visual so you could

58:27 python based visual so you could leverage them there’s definitely some limitations there but you could use those libraries as well inside desktop there’s Automation and integration python can help you automate things and then lastly there’s a this is what the point you were making Tommy was Community Support python has a large and very active Community with abundant resources tutorials and forums available so odds are you’re going to find some answers I really like that answer because I think that’s one of the strengths of that language is it’s been around for so long it has so much General help it really makes easy so

58:58 General help it really makes easy so it gave me a little in conclusion in conclusion while it’s not necessary to Learn Python for power bi development it can enhance your capabilities and offer more flexibility in data manipulation Advanced analytics custom visuals and automation consider your specific requirements and goals to determine whether investing your time in lithon would be beneficial for your power bi projects projects it didn’t talking about cloud origin or engineering but I think it a lot of the main points here it talked to

59:28 lot of the main points here it talked to the strength of python and I think that’s that’s a very good starting point but it was it was okay so I would I guess we gotta I guess we gotta score it again at this point so what’s your score Tommy how did how did chat GPT go yeah it’s a solid seven out of seven for me oh all right Seth how about how about you that all boxing I I don’t give tens oh yeah it’s like that’s Perfection that’s Perfection it’s like a 8. 9 out of 10. all right solid I’ll I’ll give it a I wouldn’t probably give it a

59:58 give it a I wouldn’t probably give it a full seven but I would probably give it a six out of seven okay I’ll give it a success I’ll stick with Tommy scale the odd the odd number there I won’t go to a thousand like I usually do you’re seventy six five yeah 761 out of 1000 that’s that’s six thousand out of the 6100. excellent well thank you very much for listening to podcast we appreciate your listenership thank you everyone who was in the chat window amazing conversation that was going on so thank you so much for chat it was super fun listening to everyone talk and discuss things

60:29 to everyone talk and discuss things everyone was listening to Greg as he as he wields his data stack knowledge which is awesome I love it’s literally like we have like a Wikipedia of data knowledge that that is Greg so Greg thank you very much for participating and adding tons of really great feedback inside the comments that is super fun I really love enjoying and listening to those our only ask with the podcast is if you like this content if you like what you’re hearing here please share it with somebody else on social media on someone to work just let them know you found an interesting podcast that’s helping you think through or have those water cooler conversations around power

60:59 water cooler conversations around power bi and data things Tommy where else can you find the podcast you can find the podcast anywhere they’re available apple and Spotify Google podcast make sure to subscribe on those audio platforms you join the conversation live every Tuesday and Thursday 7 30 a. m Central on all power bi tips social media channels excellent and remember as this episode this episode summarizes the the roach’s maximum which is transform your data as far Upstream as you can and as far Downstream as

61:29 you can and as far Downstream as necessary necessary I feel like there should be another ending phrase but that’s that’s all I got right now that’s what we’re gonna get today so thank you all very much and we’ll see you next time

Thank You

Thanks for listening to the Explicit Measures Podcast. If you enjoyed this episode, please subscribe and share it with your team.