Advice to becoming a Data Analyst?

[Acorns844]

Hey guys

I am currently in the 3rd year of my Bsc Applied Mathematics and Computer Science at UNISA .My field of interest is Data science and/or Embedded systems.

My question is what do i need to know and be able to do ,to be able to apply for Junior roles in Data science specifically.Should i be learning SQL,Python and PowerBi? Will this be enough to land me a job in the field?
Which projects are advantageous and will help me stand out ?

My second question is would a company hire a final year student for a Junior role even though i haven't completed yet?
I really want to increase my chances of employability so any help is good.

Hoping @cguy can chime in as i see he is well versed in the field.

 

[Pegging]

All the above. Also enjoy having a good head of hair while you still have it.

 

[B-1]

Studying and trying to prove yourself as a junior in a tough field is going to be very hard. If you can stick it out til at least your last 6 modules I would recommend that.

Your degree is a great start but the next question a typical hiring manager is going to ask is how long until this junior is productive and not a drain on other resources.
You already listed SQL, Python and PowerBI which are great for showcasing actual ability so get those going and then start building some real world analysis you can use to get your foot in the door.

Find a public dataset online, create analysis that would be of strategic, marketing or other value and present it in a way that would be easy to analyse. Then do it again until you have enough to get an offer.

 

[cguy]

In the areas that I've worked in, the core toolset is usually Python, Numpy, Pandas/Polars, etc. In terms of academic skills, maths, applied maths, statistics, and computer science are very useful. Depending on direction, machine learning, deep learning, PyTorch, etc. could also be useful.

In SA, I've heard that things like Power BI, Tableau, etc. are often used. I've never encountered them though.

I wouldn't try for a full time hire, but intern roles for undergraduates are very common (outside of SA, pretty much all interns are still studying). It's a great way to get experience.

Another consideration is what type of data science you are interested in: finance, research, operational research, medicine, etc. Getting a background in the domain you want to go into is generally something that will give you an edge.

Embedded systems is very different of course. What do you lie doing here? Bare metal CPU? FPGA? Microcontrollers?

 

[Supine]

Consider doing some Kaggle challenges to add to your portfolio too.

 

[Acorns844]

Embedded systems is very different of course. What do you lie doing here? Bare metal CPU? FPGA? Microcontrollers?

I would need your guidance in that as i only have microcontroller exposure when it comes to embedded systems.Do you think its worth it trying to learn embedded systems in south africa?

I wouldn't try for a full time hire, but intern roles for undergraduates are very common (outside of SA, pretty much all interns are still studying). It's a great way to get experience.

Thats my issue right there , in other countries they take final year students as interns but in SA where we live they do not ,hence im asking for what a junior should come in knowing?

Another consideration is what type of data science you are interested in: finance, research, operational research, medicine, etc. Getting a background in the domain you want to go into is generally something that will give you an edge.

My goal is to work in a Bank/Finance related role when it comes to data science.

 

[Vengeful]

Head of Data and AI for the largest Mining Services Org in the world here <-

Programming is becoming less relevant in this field.

We use Palantir and the interfaces that are pre-built into the next gen platforms have already the tools you need to clean, sort, build pipelines, initiate machine learning and even with its embedded AI optimise the algorithms for you and do all the things my junior analysts were doing - so I transferred the lot of them to another division where their skills were necessary (finance ).

My senior team are domain experts from a business perspective - they understand the sector and can actually put together automated insights with the AI trained to act as Mining Consultants - but thats all I use them for really. The AIs can put together insights faster than anyone else can. We are also an AI first company - meaning the tools are there to use first.

We are scaling down the divisions and transferring people with the need for analytics. USA - scaled and moved; South America - I already had a lean team there; Oceana - scaled down; MEA - scaled and moved

Thats the background of the analytics and AI from an organizational level.

Now to answer your questions:

Would you get Hired?
I hired students as Data Analysts with exactly your background - I had people who like you hadn't finished their degrees yet - my hiring philosophy was to seek talent not skills.

SQL, Python, PowerBi - that's a base level - but you have to understand every other candidate is coming with this as well. To stand out, you have to show that you provided value.

One of my candidates didn't just say they knew PowerBi - they implemented it in real world scenarios - they had a side business and helped analyze their own data - which led to optimizations.

So stand out and do something different - someone mentioned Kaggle - yup I second that great suggestion - hackathons another place to look - SA has quite a few hackathons - join and learn.

I offered many internships through WeThinkCode (they were our pipeline for talent). MTN use them as well.

Are those skills enough to get you hired?

For an Intern fixed term contract - Yes it is - because I will scale you up to where you need to be.
Then from intern after contract - you can be promoted to Junior.

We revised our entry level position requirements as well ( I have since added - AI generation of insights - and I have added a Case Study to the hiring process - testing analytical thinking of a real world problem and framing of questions to AIs)

PowerBi isn't enough - I would suggest learning Fabric (Microsoft is pushing heavily for all companies to transition) - Fabric is a full turnkey solution - you will be more valuable in the market if you know the full set of Fabric tools

SQL - For us the Ontology is designed and AIs have learned them - so SQL isn't important anymore for us. BUT that being said being able to understand how to put together data from Data lakes is important - nothing beats a quick SELECT * from at times

Python - the blocks for your latest and coolest ML algo is already designed - drag drop connect inputs and you are away - knowing which algorithm to choose is more important. So I saw someone suggested learning about the WHY rather than the HOW - love that. I second that comment.

Ultimately - the future of where I see Data Science being - is that its about understanding the business problem needing to be solved - framing the hypothesis and asking the AIs to run the tests - then the results of those tests are built into Production ready algorithms.

 

[cguy]

You've got some good responses so far. They definitely highlight the wide coverage of the term Data Science, as well as how it can be very different in different sectors and countries.

I would need your guidance in that as i only have microcontroller exposure when it comes to embedded systems.Do you think its worth it trying to learn embedded systems in south africa?

I've been removed long enough, that I can't really comment on embedded in SA. Overseas there are definitely opportunities, but the far more lucrative tangentially related options are FPGAs and HPC (i.e., not embedded as such, but digital logic design, and performance software with strong tie hardware). The demand for these are huge.

Thats my issue right there , in other countries they take final year students as interns but in SA where we live they do not ,hence im asking for what a junior should come in knowing?

So if finance is where you want to go, make sure that you have significant (relative to other new grads) of the subject, in particular, get into the details if there's something you're particularly interested (derivatives pricing, modelling, time series prediction, algo trading, etc.).

Also, don't limit yourself to just local opportunities. If you can show initiative (projects, articles, Opensource contributions, etc.), you may stand a chance of getting in somewhere overseas for an internship. That would be a huge win for you.

My goal is to work in a Bank/Finance related role when it comes to data science.

Banks and finance in general have a lot of opportunities here, so it would be good to narrow things down a bit.:

You have data engineers, working on the data problems (correctness, quick availability, optimal layout, storage, data interfaces, etc.).

Then you also have QRs (quantitative researchers), who work on things such as derivative pricing, foundational analysis (investment decisions based on company and industry reports, macroeconomics, etc.), machine learning based timeseries based price/portfolio predictions, and trading strategies.

There are also QDs (quantitative developers), who work on the software stacks used by QRs. This can include computational libraries, research and trading platforms, optimizing mathematical code, etc. With QDs, this often also means using advanced knowledge of things such as maths, approximations, precision requirements, etc., to make tradeoffs between optimal theory vs tractability, computational costs and latency.

Generally speaking, QRs tend to get paid more than QDs, who get paid more than data engineers.

 

[SBSP]

Knowing how to use tools like Python, Power BI, and SQL is important, but what is often overlooked is the industry in which you plan to apply data analytics. In many cases, domain knowledge can be just as important. If not more important than technical skills.

For example, if you want to apply analytics across multiple modules in an ERP system, it is worth investing time in understanding how ERP systems function, how the modules interact, and how data flows between them.

The same applies to other domains such as warehousing, logistics, sales, retail, rentals, and similar industries. Each operates within its own processes, terminology, and business logic. Without understanding that context, even strong technical analysis can miss what truly matters.

Ultimately, if you clearly understand what you are trying to accomplish, much of the analysis can be done in Excel without any programming languages. Technical skills become essential after you have defined and discovered what needs to be done, particularly when you want to build repeatable processes, scale the solution, and introduce automation.

 

[FiestaST]

Besides all the "common stuff" these days they want end-to-end thinking with business acumen.

What business problem and/or solution can you provide/solve?

Additionally the ability to leverage AI to work smarter & more efficiently.