Edition #10 Three mental frameworks for problem solving; taking a ride in viz
🧠 Mental framework
Mental frameworks are design patterns for daily life. They are tried-and-true methods for problem-solving. In this edition, we will unpack a problem-solving toolbox with 3 powerful mental frameworks.
Why we need these tools in the first place? The problems we tackle tend to fall into 3 types:
problems with known solutions
problems which the solutions and the existence of solutions are unknown
problems which the questions are yet to be clearly defined
For the first type, having solely the technical know-how would be enough. But for the next two, one needs something beyond.
⭕ First principle thinking
a.k.a. think like a physicist
Focus on the essential question, and not be distracted by any derivatives. First principle thinking is about reducing the problem to its irreducible core. It’s about putting aside assumptions, putting aside analogy, and see the problem afresh.
🤿 Second-order thinking
a.k.a. think like an investor
After the initial solution, think about its side effects and long-term consequences. This is because the first thoughts that come to mind are often the superficial solution anyone can arrive at and may lack finesse. Second-order thinking is about diving beyond a straightforward solution and bringing in other people’s alternative views.
💠 System thinking
a.k.a. think like a policy-maker
System thinking is about going beyond linear thinking. It assumes unexpected/undesirable outcomes might happen given our connected world. One of the visual aids for this framework is the connection circle, which helps identify the cause-effect feedback loops, either reinforcing feedback or balancing feedback.
Relevant reads
30 mental models to add to your thinking toolbox
🧉 Worth viewing
360-degree rollercoaster viz in R by Tlyer Morgan-Wall
🔦 Kaleidoscope
As we are nearing the end of quarantine, here’re some creative+code project ideas to fizz up indoor living.
In case you are thinking of making a fashion statement using CircuitPython.
Tools required: LEDs, micro-controllers
Or perhaps instead of painting with light, you prefer sculpting it.
Tools required: a projector and some pieces of paper.
Or one can make motion graphics by playing with ASCII in Rstudio.
Tools required: a video recorder.
📚 Recent Reads
Machine Learning Design Patterns: Solutions to Common Challenges in Data Preparation, Model Building, and MLOps by Valliappa Lakshmanan, Sara Robinson, Michael Munn
This book won’t tell you how to frame and approach ML problems. Instead, it helps with the myriad of issues you may encounter during the actual execution, from data representation to reproducibility and serving. It lists those suggested solutions (which can save time re-inventing a solution to a common problem) together with the tradeoffs.
I also learned that the phrase ‘design pattern’ originated from an architect Christopher Alexander half a century ago, who first devised the term to describe re-usable solutions to all kinds of recurring architecture problems.
Hold Your Dreams to Test by John C. Maxwell
I found a few gems from it relevant for people aspiring to get into data science.
Check if your dream is your dream.
This brings about the fundamental question of validating your dream. How do we know that the dream is an internal calling, instead of an accumulation of external influence from media, culture, and peers? The book suggests a litmus test – ' the fulfillment question’ by checking if working towards the dream brings satisfaction.
Think about what you're willing to sacrifice, and what price is too high.
Working as data professionals requires constant up-skilling, and many have put in lots of effort and time. The fact that the role is broadly defined is only adding to it. The book suggests making a list of things you would never want to compromise (e.g. health etc).
Also,
Think about who will your dream benefit.
🍸 Jamais Vu
‘Never seen before’ - the opposite of déjà vu. This optional section contains either 1) new concepts I’ve freshly learned or 2) terms that I was familiar with but started to view in a new light. All concepts are fluid. Best served with a pinch of salt.
Goodhart’s law
When a measure becomes a target, it ceases to be a good measure. Mentioned in the book The Tyranny of Metrics. This rings true in many of the data science work. A metric, once known, can often be gamed or supersede the reality with the distorted proxy.
Shotgun debugging
(The bad practice) of making relatively undirected changes to software in the hope that a bug will be perturbed out of existence - from definitions.net
Cobra effect
Well-intended incentives can have undesirable effects. It is based on the story where the British government offered a bounty on cobras in India (during colonial times) in the hope of eliminating the venomous animal, resulting in people breeding cobra for profit. The term was coined by economist Horst Siebert.