Become a fan of Slashdot on Facebook

 


Forgot your password?
Close
typodupeerror
×
Education Microsoft

Microsoft and Jeff Bezos Tap Excel, Not Python Or R, To Teach Kids Data Science 188

Posted by BeauHD from the R-you-serious dept.
theodp writes: Are you ready to rock it with #datascience?" asks a tweet from Club for the Future, the tax-exempt foundation founded and funded by Jeff Bezos's Blue Origin, which is partnering with Microsoft's Hacking STEM to show how data science is used to determine a Go/No-Go launch of a Blue Origin New Shepard rocket. Interestingly, while Amazon founder Bezos and Microsoft CEO Satya Nadella are big backers of nonprofit Code.org and joined other tech CEOs for CS last fall to get the nation's Governors to "update the K-12 curriculum, for every student in every school to have the opportunity to learn computer science," Microsoft and Blue Origin have opted to teach kids aged 11-15 good old-fashioned Excel skills in their Introduction to the Data Science Process mini-course, not Python or R.

"Excel is a tool used around the world to work with data," Microsoft explains to teachers who have been living under a rock since 1985. "In these activities, students learn how to use Excel and complete all steps of a mission by engaging in the data science process. In this mission, students analyze key weather data in determining flight safety parameters for a New Shepard rocket and ultimately make a Go/No-Go decision for launch. Students learn how to use Excel while engaging in this dynamic Data Science Process activity [which is not unlike PLATO 'data science' activities of 50 years ago]." Blue Origin last September pledged to inspire youth to pursue space STEM careers as part of the Biden Administration's efforts to increase the space industry's capacity to meet the rising demand for the skilled technical workforce.
This discussion has been archived. No new comments can be posted.

Microsoft and Jeff Bezos Tap Excel, Not Python Or R, To Teach Kids Data Science More

Microsoft and Jeff Bezos Tap Excel, Not Python Or R, To Teach Kids Data Science

Comments Filter:

  • Seems reasonable to me (Score:5, Insightful)

    by narcc ( 412956 ) on Wednesday April 05, 2023 @06:11AM (#63426910) Journal

    Excel or any other spreadsheet should work just fine.

    R isn't great, and it's a lot better than Python. I wouldn't want to use either one for a data science course for kids. They would just get in the way.

    • Re:Seems reasonable to me (Score:5, Interesting)

      by korgitser ( 1809018 ) on Wednesday April 05, 2023 @06:22AM (#63426922)
      Yeah. As much as teaching Excel in any kind of science class sounds retarded to me, what is even more retarded is the fact that kids these days seem to lack any Excel skills whatsoever. It is just the most elementary way to have the computer do some actual work for you, and seeing the young'uns minds blown by putting some numbers and formulas into a spreadsheet just makes me sad. It was the killer app of PC back in the day for a reason after all.

      • Re: (Score:2)

        by edis ( 266347 )

        It is not much even about Excel skills for themselves, as for very good foundation where computing is bound to immediate, well structured and flexible visualization of result. Similar idea was behind the Logo programming, researched in MIT.

        • Plus, it's extremely useful in real world, unlike toy programming languages. Knowing how to use pivot tables is going to be more important to 99% of the kids than how to use Pandas and Koalas which are going to be deprecated by the time they graduate anyway.

      • Re: (Score:2)

        by dbialac ( 320955 )
        Well that and the course is about data science, not programming. You might as well teach a 15 year old auto mechanics when your goal is actually to teach them to drive.
      • As a working scientist,I use Excel all the time. For case where there is lots of human data entry, modest amounts of data, and not a lot of calculation, it can be a great tool.

    • Re: (Score:2, Interesting)

      by Anonymous Coward

      Yeah I think it's reasonable, and we do need to make sure we're churning out the next generation of data scientists.

      However the timing raised eyebrows a bit for me given data scientists seem to have born the brunt amongst actual technical roles of recent tech layoffs. If these layoffs have taught me anything it's that big tech has gone seriously cold on data science all of a sudden.

      Far and away they've been hit disproportionately. Other roles that seem to have been hit heavily are of course management, recr

      • I don't think they've gone cold on it, I think they were massively oversubscribed waiting for a future when they would have use for all those people. Consider it labor-hoarding.

        Then revenue took a hit, so they had to liquidate the warehouse stock because when you are in charge of a business, at some point you're going to have the decision to seal a hatch with people still inside or let the whole ship sink.

        Layoff suck a giant horse cock - I've been victim to more than one. But at some point in any company

    • Re: (Score:2)

      by amorsen ( 7485 )

      The problem is that there is no easy and safe way to get data into Excel. The only reliable ways involve another tool generating a .xslx file or using a database link in Excel. Neither are good options for teaching 11-15 year olds.

      • This is a process design problem. They will likely be taught a tedious import process from a delimetered data file. As the process design issue you raise is effectively why no real world operation like this would use excel.

        This doesn't negate the experience though because designing some front end to load a genuine data stream would effectively include many of the same basic criteria and checks to make the this ultimate "go" or "no go" call. It's just the real application is more robustly designed.

      • The problem is that there is no easy and safe way to get data into Excel.

        CSV import works fine. There are tools to clean up CSV data before importing it. The problem with importing data into Excel is that it only handles 1.04M rows, which is very few records by modern standards. I was going to use it to clean up and analyze some CSV files before importing them into a database and couldn't, because it wouldn't handle enough rows! So I wound up writing Perl to do it.

        For a kids' data science class, the CSVs can be short and preformatted so they don't have problems. All they're like

        • CSV import works fine...for kids learning. It chokes in so many ways on seemingly basic data.

          Generating Excel programmatically is not as hard as it used to be. It's generating an XML file and putting it into a folder structure before zipping it. Which is needlessly complicated but at least it's not an undocumented binary format.

          • CSV import works fine...for kids learning. It chokes in so many ways on seemingly basic data.

            Well and IRL you can use a database, OLAP or relational. Or the data will be already in an xlsx file because that's where all the business stuff lives.

        • If you're 11 years old and still learning the basics, you don't need 1+ million rows of data to do that. You need maybe 100, because you're learning the logic processes and the amount of data being shoveled through is immaterial as long as you also teach that while spreadsheets are a convenient tool to use for learning the concepts, nobody that wants to do these things at any kind of useful scale would use a spreadsheet in favor of the more complex and far more scalable tools out there.

          Trying to teach a cl

      • Re: (Score:2)

        by stripes ( 3681 )

        Wait, CSV files aren't "safe"? Sure they are awkward to deal with if the data naturally has commas or quotes, but many kinds of data are pretty light on that sort of thing. To be fair I don't really work with spreadsheets very frequently, and it may have been a decade or two since I've needed to get anything into Excel in particular, but I remember it being "not hard" to pull data out of a "real" database, massage it lightly with Perl, and produce a CSV that excel would accept. Likewise I remember being

        • No, they are not safe. Tons of data exists in the world that contains characters that CSV files can't handle. Like commas. I have a vendor that insists on CSV files, and they sent me one just the other week which had commas in address lines.

          You can imagine how well Excel handled that.

          In addition, CSV can't handle metadata, and lots of data needs that extra information to interpret it correctly.

      • If only there was some simple, text based, format that allowed for separating values with some character. Then you could import and export those character separated values to your heart's content!

        But I guess that's as much of a pipe-dream as pasting data directly into a spreadsheet.

      • Re: (Score:2)

        by Junta ( 36770 )

        It's also not a huge problem when teaching 11-15 year olds. They get it pre-imported already as course fodder. If you get into projects that actually connect data to it, then there's an ecosystem for those things to generate spreadsheets.

      • We're talking about teenagers here. They're not building an ETL process to churn through a couple gigs of data in order to crank out several thousand account interest accruals as fast as possible. They're learning the concepts, and the concepts can be taught with small data sets that are easy to copy / paste.

        Just make sure to tell them that a spreadsheet is the Hasbro My First Data Viz Product of the data science world - it's not meant for use at scale for data science. That's where tools like Python, Lu

    • Spreadsheets are probably a good enough intro to data science tool for kids, and it's good that if they try to do anything complicated with it, they'll rapidly learn that it has pathetic limitations. The number of rows permitted in a spreadsheet isn't enough for anything interesting without breaking your data up into pieces and then having to go through agonies as a result. And when spreadsheets implement significantly more rows than Excel does, they can't back it up with performance...

    • Re:Seems reasonable to me (Score:5, Informative)

      by dogsbreath ( 730413 ) on Wednesday April 05, 2023 @08:11AM (#63427044)

      Don't paint me as an Excel fanboy (far from it) but ...

      Excel is everywhere and is immediately accessible. The cell-grid structure is more intuitive than pretty much any other presentation layer I have encountered. My kids back in elementary school figured it out with just a bit of minor initial prompting; they already knew all of the arithmetic structures.

      Plus, it is THE universal program. It can do anything. Besides the usual financial planning stuff, I have seen people with no computing or programming background build:

      - a classroom lesson planner

      - an automated employee shift work planner/scheduler

      - a word processor (yup if you have a screwdriver and a file, you also have a chisel)

      - geological mining operation simulators (strip mine slope edge stability)

      - an employee behavioral risk estimator (problem gambler spotter... use of corporate credit cards)

      Basically anything that can be laid out on paper as a table with either or both text components and arithmetic calculations including calendar dates can be put into Excel. If you can imagine it on paper, transferring it to Excel is usually straightforward and requires little specialized expertise.

      Excel gives you the ability to understand and manipulate data without having to first digest a programming language. If you already have the mathematical model, you don't have to translate it into program functions and structure.

    • Re: (Score:2)

      by DarkOx ( 621550 )

      Pretty much this. If the goal is to teach data science vs do data science a spreadsheet, Excel or otherwise is a great visual tool to get ideas across.

      Numpy is a great library - but that isn't really the point here. Python is a general use interpreted (yes yes you can compile it) language that frankly is 'abused' a lot for data science work. It gets the job done especially where you have a common case of a handful of data science people directing a larger number of software people. Having the tool oriented

    • Exactly, this seems like someone just wanted to point out how Microsoft is cooking for the home team, but in reality a tool that allows you to visually see what you're doing with data transformation and be able to play "what if" by changing data and seeing the outputs immediately change is very useful for how some people learn.

      Thinking you can take kids directly to coding in python when they don't even know simple logic constructs is setting kids up to fail. Let them learn the simple stuff with simple tool

  • way back when I said "I know how to program" (Score:5, Insightful)

    by williamyf ( 227051 ) on Wednesday April 05, 2023 @06:15AM (#63426914)

    So, "whatevur" people can do in Excel, I can do better in C or Pascal...

    I lived to deeply regret that youthfull arrogance...

    So, this seems like the right move, teach the kids data science with the most useful tool for other areas, and let them self-select to learn the more powerful ones...

    • So, "whatevur" people can do in Excel, I can do better in C or Pascal...

      I challenge you to sit down at my work machine and do better than me at Excel. Hint: The machine is locked down, you don't get to install a C or Pascal compiler. You may have some luck with VB, but C or Pascal is absent from 99.9% of the corporate world meaning that what you're good at is utterly meaningless in practice.

  • Help kids learn the concepts. There are many tools (some better than others) that can be used as part of that learning journey. Over time they will figure out which tools best match their needs.

    • Help kids learn the concepts. There are many tools (some better than others) that can be used as part of that learning journey. Over time they will figure out which tools best match their needs.

      By learning the concepts they can then apply them with other, more powerful tools. Way back when, I learned how to do a half interval search in FORTRAN. I've not used FORTRAN in years, but applied the search concept a number of times. Best uses was on teh GMAT where I could quickly answer match questions without ever really solving the problem.

  • Obviously, Microsoft is going to promote Excel, but Bezos?

  • Kids won't learn any data science from excel. (Score:5, Insightful)

    by oldgraybeard ( 2939809 ) on Wednesday April 05, 2023 @06:32AM (#63426926)
    What Microsoft is teaching is product lock in.

    • Excel gives you robust, easy visualization tools, and learning is often about being able to express information in different ways. The master lesson is when the easy tool is no longer the right tool.

      • Excel gives you robust, easy visualization tools, and learning is often about being able to express information in different ways. The master lesson is when the easy tool is no longer the right tool.

        You hit the nail on the head.

    • Yeah, because simple Excel concepts aren't exactly the same in Google Sheets or LIbreOffice Calc, are they?

      I think even Apple's my-first-spreadsheet thing they ship for free with Macs and iPads use exactly the same formula syntax and function names.

      When you need to wipe your nose any facial tissue will do, even if you persist in calling it Kleenex.

  • FOAD (Score:3, Funny)

    by Anonymouse Cowtard ( 6211666 ) on Wednesday April 05, 2023 @06:32AM (#63426928) Homepage
    It's Google Sheets all the way, cunts. On Android and in the browser on the desktop. For my homebrew, for specing out components and prices for PCs, for tracking my listening habits with data from librefm and on and on. Excel... Pfffft

    • Once upon a time I'd have grumbled with Google sheets being used for anything. But its actually pretty solid these days, and dear fucking lord the CSV export on excel is mindbogglingly broken. Shits the bed if you have more than one tab (what happened to just asking what tab(s) to export, and it outputs it in some whackjob Unicode format that literally is different depending on what country your in.

      Horrifyingly useless. Google sheets, Libreoffice calc, Apple Numbers, all great options and actually export n

  • ...But Python and R are ... use the expensive business tools to learn data science ...

    • Re: (Score:2, Informative)

      by thegarbz ( 1787294 )

      Excel is free. All Office online webapps are free to use when you sign up to a free Microsoft account. They are also free to use on tablet devices. Only Windows applications require you to pay. Also Office is free for everyone in education.

      The skills learnt are transferable to LibreOffice's Calc.
      The skills learnt are actually relevant when you enter the workforce where you will almost certainly be using a computer with neither Python or R installed.

  • They'll be better prepared this way (Score:5, Interesting)

    by DeplorableCodeMonkey ( 4828467 ) on Wednesday April 05, 2023 @06:52AM (#63426948)

    Most of what businesses have traditionally done that could be called real world "data science" has always been done with Excel anyway. Heck, most of the field would die in a fire if Microsoft ever created a Trifacta-like big data backend that seamlessly integrates with Excel so that you could "do Excel at big data scale" for things like loading TB-sized spreadsheets and doing joins. It's been an unmerited grace to much of the industry that Microsoft has apparently overlooked this opportunity.

    • last year I had an engagement where the business wanted to drive Excel-generated analysis with data in Snowflake. It was laughable and after a couple of days using Sigma they were striking their Excel flags. Sometimes you need a freight train instead of a truck especially if you're dealing with millions of data points.

  • Not every 11-15 year old is a scientist (Score:5, Insightful)

    by thegarbz ( 1787294 ) on Wednesday April 05, 2023 @06:56AM (#63426950)

    Being able to use Excel to analyse data is a far more relevant skill for school kids. Leave R and Python to the university students who chose a career path that will actually require those kinds of tools.

    • Not sure about that leaving Python to university students. Teaching Python and even 'C" for robotics is increasingly common and Python is an easy language for children as young as 7. If we want our kids to be ready to be competitive, they should be at least familiar with these tools. MIT's Scratch ( https://scratch.mit.edu/ [mit.edu] ) has been used to teach programming young children for over a decade now and Python seems the next logical step for basic computer knowledge. If kids are using web enabled gaming consol

  • Excel is a limited tool and "End User Applications" implemented in Excel are far more a problem than a solution. It is a hallmark of these "teach kids something with coding" that 2nd rated tools are used, because the whole thing is a PR activity anyways and has no actual value besides that.

    • When it comes to data visualization Excel is actually often one of the best tools out there. To this day I can generate more useful graphs in Excel than anything else. If visualizing is not a priority then of course its value drops off.

      • Re: (Score:2)

        by gweihir ( 88907 )

        The claim was "data science". That is a bit more than just pretty plots.

        • The story was about teaching data science; visualization is actually a big part of that; people need to be exposed to information in different forms to fully absorb it.

          Many times a graph is much more than a pretty picture, but a way to see how events are correlated. This is especially true when you can dynamically create trend lines or graphing deltas and integrals alongside core data.

        • The claim was "data science". That is a bit more than just pretty plots.

          But it's also not writing boilerplate to glue Python libraries together, either.

    • I agree that production apps implemented in Excel are a nightmare. I disagree that Excel is a suitable way to introduce kids to data manipulation and transformation.

      • Re: (Score:2)

        by gweihir ( 88907 )

        Oh? And what tool do you think these kids will be using when they have grown up and like to implement some automation bypassing the IT department? Right.

    • Excel is a limited tool and "End User Applications" implemented in Excel are far more a problem than a solution.

      Depends on who you ask. Actually, they've facilitated vast amounts of business. If they weren't a pretty reasonable solution, they wouldn't be much of a problem. They become problems for other people, not the people who built them, or who use them. Don't minimize their utility, or their impact.

      Even today, have you looked at the cost of developing "real" software? I, like many others here, are involved in that business for a living, and we see the costs. When a company with limited resources, when presented

  • April fool's day was several days ago. Surprised they chose to publish this several days later. Not clear what the source of this joke was to check the publish date but no credible tech educator is going to teach Excel over Python.
  • Isn't that the defacto way to communicate in the engineering world? Create a PowerPoint deck and bore your audience to death?

  • This is how Microsoft has asserted dominance (Score:5, Insightful)

    by ClueHammer ( 6261830 ) on Wednesday April 05, 2023 @07:28AM (#63426982)
    For a long time. They program the kids to need Microsoft products so when they get to be adults that is what they know. Data systems should be platform independent. not bound to Microsoft only products.

    • That's definitely true. But it is also a concept lesson that would work with any spreadsheet. And importantly, most of the good spreadsheets are tied up in some ecosystem or another. Instead of learning one tool really well, there should be a pattern in education of switching up tools entirely to another equivalent product so that you're learning concepts instead of memorizing UI layouts. Especially since major vendors love to mess around with the UI from version to version anyway.

      I don't think it's a p

  • ...because excel is good enough and the tool the vast majority of people use to manipulate and interact with data today.

    I know Slashdot feels like every child should be steeped in the deep arcana of sql queries and Python, but that's just not necessary.

    • I mean... making them learn =INDEX and =MATCH seems even more cruel than SQL to be honest.
    • Yeah it makes sense to me and the progression from Excel to SQL to programming is sensible if the students want to go deeper. But plenty of scientists get by with nothing more complicated than spreadsheets for day to day stuff.
      • Then you find out "nothing more complicated that spreadsheets" actually means "with extensive macros written in a procedural programming language of some kind"
        ..and realize that you shouldnt have said what you just said.

  • The real question (Score:5, Funny)

    by smooth wombat ( 796938 ) on Wednesday April 05, 2023 @08:20AM (#63427060) Journal
    Will they be teaching kids how to use Excel to send email [unilad.com]?

    • Re: (Score:2)

      by ch_rob ( 655367 )
      ...checked my facts with ChatGPT: Excel's Hyperlink function was available in Excel in 2007, the video came out in 2002. :-/

      • ...checked my facts with ChatGPT: Excel's Hyperlink function was available in Excel in 2007, the video came out in 2002. :-/

        What did you use to check ChatGPT's facts? It makes stuff up. A lot. I often ask ChatGPT questions to find facts, but I always follow up by checking them with other sources, because ChatGPT's facts are often wrong. It still often takes less time overall to ask ChatGPT and then find confirming evidence than to find the same information without ChatGPT -- and this is often true even when ChatGPT is wrong, because it generally points you in the right direction even when the answer it provides is wrong. But yo

  • KILL IT WITH FIRE !!!

  • Excel is a defacto standard albeit it's not a tool for deep analytics like R. For example, I can't see someone doing Principal Component Analysis in Excel.
    I do see Excel being used where it shouldn't be used as the "only" data analytic tool and it's surprising the rat holes that companies build themselves when Excel
    is leveraged for things it's ill-suited for.

  • Most data analysis is done in Excel. This is mundane stuff like manufacturing timing, tracking investment performance, resource management, etc... Even if you are feeding data into a dedicated analysis tool or model, a lot of massaging is done in Excel as well. There's a reason every major data analysis platform features native Excel import.

    Maybe have an advanced class where you import stuff into R (though a lot of stuff seems to be moving to Python) from Excel, but for learning basic concepts, Excel is gre

  • Big Tech needs cheap labor. They manage that with certififications and education. Columbia exhibit A guarantees a job for every CS major teaching to exactly what industry proscribes.

  • it's an introductory level course, seems to me that many of the basic hurdles to data science can be explored regardless of the toolset (eg. badly formatted input data, relevant data mixed with irrelevant). Anything that gets people aware of what it takes to even get to the point where you can apply data science is a big boost to my mind.

  • 11 years old seems a bit too early to start teaching kids how to program in Python. While many kids have basic computer skills at point, I doubt that many of them are ready to tackle anything more complex than an Office suite at that point.

  • I don't like it, but I have to begrudgingly admit this will be more useful for most kids than a dev tool like python, R, or god forbid a purpose made/toy 'educational' program instead.

    An idea hit me the other day like a ton of bricks, and it changed how I think about product design. In this instance I think it applies to education design too.

    If you design a lightswitch that requires an average level of intelligence to use properly, that leaves half the population in the dark.

    I would have loved to have had

Slashdot Top Deals

Make sure your code does nothing gracefully.

Close