I began the MSc Business Analytics programme at UCL School of Management last September. I’ve only completed the first term on the programme, and what a journey it’s already been. It’s been challenging and frankly tough. But I have learnt so much already.

First things first. Why am I writing this? It is not just that I want to share things that (I think) matter. Though that is important, I figured that I love explaining things to others. I enjoy distilling and making complex problems seem pure, lucid and, hopefully, easy, and it helps me to clarify and structure my thoughts on topics I am writing about, which means I am learning too.

Here are my key learning points from my first term studying MSc Business Analytics at UCL School of Management.

Do a sanity check

I’ve realised that it is vital to check that you really understand the data you are working with and check your numbers. Not just looking at a number and accepting it — but thinking about it. What does “increase by 300%” mean? Is it even possible? What is the benchmark? Always check whether numbers make sense to you, as for me if something wasn’t feeling right, often it actually wasn’t right.

Take a step back and ask questions

Sometimes we just tend to rush ahead and try to solve numerical problems. I, however, believe that it’s crucial to step back and think about what are we trying to achieve. Why a particular model is needed? Is it the only way to get what is needed? Are “numbers” representing the “concepts” we are trying to grasp? Building a model and predicting labels is actually not that difficult. Asking the right questions, collecting appropriate data and setting a sound logic — is.

Catch that low hanging fruit

In a strategy and consulting world, there is a concept of catching the “low hanging fruit”. That is, you can make big improvements by doing something relatively simple. I believe this theory applies when working with data. We often overthink things, which I did when trying to build a sophisticated model like a random forest, SVM, neural net etc. But I’ve realised, starting with a simple and easy model (say linear regression) usually works. And, you can build layers of complexity later on.

Not too techy for pen and paper

With modules centred around programming and data analysis, I thought would hang up my pen and paper and switch to staring at lines of code on a screen, and I even actively tried not to write things down the old fashioned way, as it didn’t look “techy enough”. But it turned out to be quite the opposite. A lot of thinking (that matters) happens before you import a “scikit-learn” with “pandas” and start wrangling your data. A lot of my pre-processing thinking happens on paper. Whether it is how to merge my data, which kind of decision tree to build or explore casual relations — I am fond of sketching things out. It works for me. I am not saying everyone should do the same, I am simply saying that: if it works, do not be afraid to utilise that.

Python is a language

Python is a programming language and for me learning how to master it should be the same as learning a foreign language. It is essential to keep thinking of it this way and try to speak it. We always make this mistake in the beginning, when we start learning a new foreign language, or at least I did. I tried so hard to memorise all the structures, grammar and words. Which is necessary, but it’s much more productive and enjoyable to feel a language, to talk and to joke with it. Thus, I think it is important to start “talking python”. Sometimes you would not know a particular syntax, but you can quickly figure it out if you know where you are trying to get to, where to search and how to adjust the code. If you “talk python” you can also guess a syntax, just like you do with English.

Thanks for reading my first article. I hope you found it useful for an insight into the first term studying MSc Business Analytics at UCL School of Management.