Transforming and visualising data using Power BI

The past two weeks I was introduced to the ins and outs of Power BI. Four full training days I’ve been practising doing transformations on columns, making calculated measures and dragging columns and measures into visualisations. For those who are not into data analyses, Power BI is a piece of software developed by Microsoft to handle data sets. When spreadsheets are no longer sufficient to handle your data, you can step up the game by using Power BI.

Before this training I practised with SQL and Python to create scatter plots and calculate summations, and I have to admit that after using Power BI I finally understand what kind of actions I was doing to data sets when using Python. Power BI is a visual tool, so you click on the transformations you need to do to prepare your data and the results are immediately visible. And you can easily undo a step with one click.

I wouldn’t say Power BI is data analysis for dummies, because you still need to know conceptually understand what you’re doing to the data, but I totally see why many people prefer using Power BI over messing about with Python. It is visual, quicker and can create interactive reports and dashboards. The reporting part is (for now) least interesting to me, as I don’t work in a big company with lots of (sales) data that needs to flow through the organisation. However, I do feel more confident after the past weeks that I’m capable to get meaningful information from data sets. And that was the whole point of investing in this course.

Door |2021-05-04T14:26:14+02:004 mei 2021|datascience, flow|0 Reacties

I passed an exam

If my memory is correct the last exam I took was in 2004, when handing in my master thesis (on blogging and Habermas, when blogging was still new and shiny). That was an oral exam, for two of my professors. I really can’t remember the last paper-based exam I took before being allowed to hand in my master thesis. It probably was not a memorable subject or one of those mandatory statistical analysis exams. Since 2004, I never needed to sit an exam for anything. Not even for an assessment for hiring purposes, as I’ve been self-employed since finishing university.

Today I broke that examless streak.

The program at Techionista is thoroughly sponsored by Microsoft and therefore I’m learning all about Microsoft Azure. And to be able to learn that you don’t just read documentation, you increase your knowledge by practicing for an exam. Today I took my first exam, on the Azure Fundamentals (AZ-900 for insiders) and passed it with a proper score of 820 (700 needed to pass).

To avoid installing proctoring software on my computer I reserved a slot at the nearest test center. That happens to be in my home town and I learned later that it’s run by an institute that teaches IT skills to (young) people who are either on the autistic spectrum or highly gifted (many of whom can’t manage to fit into the standard school system and drop out without a degree). I noticed that the person who took me through the sign-in procedure made sure every rule in the procedure was followed in a kind manner, he properly guarded the silence in the hall next to the exam room, and as a bonus earplugs were available for all examinees. I’ll schedule my next two exams here as well.

Door |2021-04-12T12:28:54+02:0012 april 2021|datascience, flow|0 Reacties

Vrouwen blijven onzichtbaar

Het is natuurlijk absoluut simpel om te registreren of een deelnemer aan een klinische studie fysiologisch man of vrouw is. Als je dan een nieuw ontwikkeld vaccin mag testen op pandemische schaal zijn de getallen ook best snel statistisch significant. Hoe logisch is het dan om ook even een kolom m/v op te nemen in je database van gerapporteerde bijwerkingen? Ik dacht dat medische onderzoekers allemaal wel een kopie van Invisible Women op hun nachtkastje hebben liggen. Wat een naïeve gedachte van mij zeg.

[…] de vaccinmakers hebben het element ‘sekse’ goeddeels genegeerd in het vaccinonderzoek en de behandelmethoden van Covid-19. Zo had geen van de gepubliceerde klinische proeven van vijf coronavaccins de opgetreden bijwerkingen uitgesplitst naar sekse.

Hoe vrouwen vergeten werden in het Covid-19-onderzoek (bron: Trouw)
Door |2021-04-08T21:14:02+02:008 april 2021|datascience, vrouw|0 Reacties

Becoming brainwashed by MS

A big part of my data & AI course is getting to know Microsoft Azure. This week I started learning the MS fundamentals course (part 1 – 6). It’s a crash course on server terminology, such as virtual machines, containers, VPN gateways and virtual networks. It’s a lot to take in and I’m not sure how much use I’ll have of the ins and outs of Azure, nonetheless it gives me a better understanding of it all so I can become a better translator between real server and data nerds and those who are not.

Door |2021-03-09T19:16:30+02:009 maart 2021|datascience, flow|0 Reacties

Learning about learning

Now that I’m a student again, and subject to a substantial online course which has some elements that feel to me like a waste of time, I got curious about the latest insights in how we learn best. I spent my first year at university studying educational science and I still have some fascination for the design of courses and workshops, both online and offline.

The element that I found (very) inefficient was watching a teacher code in SQL for a full day, without much interaction other than asking questions via chat. During that same week we got assigned three modules to practice with SQL in Datacamp, starting from the very basic level and then moving on to more complicated stuff. I felt a bit frustrated with this set up as it seemed to me like first spending a day learning nothing, before the real work could commence in a week where I was already strapped for time.

So last week I spent the whole of Monday watching the teacher. During the morning the speed of all the basics felt too slow. By 14:30 my mind switched off, my body started yearning for movement and fresh air, and the teacher ramped up his speed to show more complex stuff leaving my mind wonder why I signed up for this course anyway. Now I don’t want to sound too critical here, as most of the course is wonderful, despite the use of MS teams to keep track of all the assignments and building a sense of community. But scheduling a session like this felt like backwards thinking, not in line with the things I know about how people learn. Nevertheless, going through the Datacamp modules went much faster than expected (and projected by Datacamp), so perhaps I learned more during that first day than I want to acknowledge.

This made me curious about what recent insights about learning are. How can I help myself over the coming months to not just tick of the assignments and move on, but actually store what I’ve practised in my long-term memory? And give some constructive feedback towards the academy?

I found some interesting resources. For instance this article published by Princeton University.

Most people believe that repeated exposure to material, such as “going over” notes, “re-reading” are the main and most important ways to learn and “absorb” information. In fact, research shows that memorizing in this way has significant shortcomings. Such methods are not only highly time-consuming and less than optimally effective, they are often rather boring. There are not only more effective and efficient methods of learning, but alternative approaches are often more engaging, interesting, and enjoyable.

How People Learn: Common Beliefs Vs. Research

Learning the basics of programming is all about doing, not memorising the exact code. That’s the message my academy sent out. Rightfully so. However, it is important not to effortlessly sail through the assignments. The article also states that “effortful learning usually signals not only deeper learning, but more durable long-lasting knowledge.” I found the SQL modules in Datacamp easy, as most of the challenges were heavily pre-scripted, stating line by line what to write. That could mean that I learned less than I did with the Python modules, where I sometimes was struggling with some concepts. Fellow students report many more issues in grasping the material and take twice as long to go through the assignments. They might be learning more and better than I am as it’s more challenging to them. I therefore will try to do some of the extra challenges in Datacamp to up the “desirable difficulties” for me.

Rushing through the modules might not be the best strategy either. It is important to “interleave” studying, or leaving space between study sessions. This can be a bit hard to do given the fact that I need to finish certain modules within a week. Nevertheless I could try spreading the work between the days and within the week. If the mandatory online sessions allow me to.

There are more principles in the article, but some of them are less applicable to the practical data science skills I’m learning right now. Or simply impossible due to current restrictions like varying locations where you’re learning.

Elsewhere I read it is important to practise extensively. “overlearning reduces the amount of mental effort required, leading to better performance”, according to this article.

According to that same author the key to learning is knowing how learning works. I’m well on my way to learn more effective then.

One other resource I’d like to mention is The Science of Learning. This is a more practical guide to translate cognitive principles to classroom situations.

None of the articles I read mentioned watching a teacher do something on a screen for a full working day as a good teaching strategy, though. I’m glad it was only one day, not the five day training the teacher mentioned doing too. I guess I just found writing SQL statements easier than writing Python.

If you have any good recent references on learning, please share them with me in the comments. I always have an appetite for learning more.

Door |2021-02-16T18:21:03+02:0016 februari 2021|datascience|0 Reacties
Ga naar de bovenkant