Thought, Thinking, Think
This week, I learned how to be more organized. And— spoiler alert— it’s not about getting my room tidy or using Notion. It’s data analytics.
It’s such a pleasure to study with different perspectives all the time. Especially when you’re in tech. When I first learned about coding, my eyes were always fixed on how to make the code work, then I knew that a working code is not all that. I also had to make it run efficiently. And also clean. Then with design, aestheticism was only just a bit of it, there’s this about user experience: understanding the interaction between human and system bridged with design.
And then with data, I had drowned myself in techniques: statistics, tools, algorithms… And yet I often forgot how to start with thinking. When given data — all raw numbers and text — I tend to look at it and wonder: “What kind of algorithm will make sense of my data?” or “How should I clean this?” and then without knowing what I was going to do, I wondered, “what kind of graphic will make it more interesting to be read?”. Now thinking about it, how annoying it is to work without purpose or guidelines!
Here’s 3 steps that will help us to be more organized on data analytics!
Meditate
The idea of meditation is to stop the mind rushing in an aimless stream of thoughts. So before going all crazy with the techniques, we should take a moment to create a purposeful process. Always question before starting to move on to the whole process of data analytics.
What is our problem?
Okay this sounds weird somehow, but really, what is the problem that makes us wanna go all the way to work with the data? Do breakdown our problems with Root Cause Analysis and Issue Tree. In a nutshell, working with finding the root problems will help us frame out goals too.
Read more about Root Cause Analysis and Issue Tree
By knowing our problems, we can create hypothesis. Having hypothesis will help us underlying theory, measure the validity and reliability, and focus on the outcomes — goals — of our analysis.
What are the measurements are you looking for?
When fixing a problem, you might want to know how to measure your problem and how to find the solution to it.
Assume we want to find out why there’s a decrease at total transaction in our platform, our goals is — of course — to increase the transaction back or even higher, and with all this we can focus on the most obvious metrics: Total Transaction.
So from there we can finally think “what data do I need?” and “Which metric do I use?” then move forward with working with the data.
Interrogate
“Life can only be understood backwards; but it must be lived forwards.” ― Søren Kierkegaard
At this point I think I just made an excuse to plug in my most favorite quotes. But again, data was what happened, and understanding data is actually a form of going forward.
I think this one relates the most with processing the data and preventing producing a whole garbage. Cleaning, transforming, fitting the model, and all that stuff will be pointless if we don’t interrogate the data. Interrogating the data can be started by doing Statistical descriptive and Statistical inference.
Statistical descriptive helps to provide information about variables in our data and highlight potential relationships between variables. Some Statistical Descriptive that can tell more about our data:
- Frequency (how often a response is given)
- Central tendency (how an average of most commonly indicate response)
- Dispersion or variation (how spread out the data are)
- Position (compare scores to a normalized score)
While statistical inference helps in the process of drawing conclusions about populations or scientific truths from data samples like the population (also by testing our hypotheses). And be responsible for every method we use on creating the conclusion.
Narrate
“The only thing to do with good advice is to pass it on. It is never of any use to oneself” — Oscar Wilde
It’s so meaningless to find so much but can’t tell that much. Best deliverance we can give with all our findings is to narrate them so more people can understand it. Narrating it with an analysis document is basically wrapping up all your process into one well-ordered writing. I think personally, this might be a step to evaluate all the steps too.
Make sure to tell the process with a story: what has been done or what will be done. State the data sources, the transformed data, what hypotheses we use. and what approach applied. Some tips on writing analysis docs I learned were being specific and detailed about the process, and never being biased by your limited knowledge. Knowing that you don’t know everything is actually so helpful to widen your mind, especially on inferring the result and giving recommendation.
Reflection
So… that’s some of my favorite things I learned this week!
When learning, things might be tricky since I have to constantly shift my thinking, and the more I know, the more I know I don’t, right? So often, I see all the process, the new knowledge, the projects, as something that I need to devour on slowly. And by writing, I reflect all of those back again. I like to make myself believe that I am nowhere knowing everything. And with all that, I can make more spaces for new things and new perspectives on the way.
In the future, I hope I can be wiser working with data.
**Note: I use several paintings from John Lavery in this article. No particular reason. It’s just beautiful, isn’t it?
