Tales from a Career Pivot — and we’re off!
I’ve decided to chronicle my journey retraining to become a data scientist. The rest of the series can be found in my profile page.
Bang! The starter’s gun has sounded and I’m out the blocks. The opening bend is the secondary school algebra curriculum on Khan Academy.
To gauge my existing knowledge, Khan Academy creates a series of ‘Mastery Challenges’, tests containing 6 questions which adapt to your answers. It feels like a rapid fire quiz. You know when their machine learning is honing in on your level of understanding as the number of questions in each set gradually reduces from 6 to 5, 4, 3 and then 2 as it narrows in on your capabilities.
The is a constant feeling of “Oh shit I used to know this! Think…..nope, nada.”
It seems I can skip the first 18% of the curriculum. Well, it’s a start.
Once you’re into the main flow of Khan Academy, micro-modules consisting of one or more demo videos plus a set of 4–7 questions progress you through their four levels of understanding:
- level one
- level two
There must be nearly 200 modules within the algebra curriculum, fortunately each day you start with a series of mastery challenges based on your recent work. I jump for joy as progress soars exponentially faster than one exercise at a time. I begin to knock off modules like a duck hunt.
Musings on education
It has often been said that if someone had fallen asleep 100 years ago and awoke today, the only thing recognisable would be what takes place in a classroom. We persist with the Victorian design of a single teacher lecturing a room of students and expect them all to assimilate the same content simultaneously.
EdTech such as Khan Academy and Renaissance Learning holds the promise of universally educating everyone well and at low cost. It is built upon the twin pillars of adaptive learning and data analytics to monitor and adjust programmes based on evidence rather than intuition.
Delivering personalised education will require data on individual’s learning style, the knowledge and skills they already possess, the goals they wish to pursue, and their progress. Educational environments can then be tailored to deliver the right mode of instruction through the most effective medium and at the appropriate pace, helping each student master one topic before moving on to the next.
I’m becoming seriously impressed with Khan Academy’s ability to customise my instruction. Once I’m around 40% through the algebra curriculum, I hit a run of very simple modules. Speed as well as accuracy is obviously one of their data inputs as I’m suddenly presented with short two question mastery challenges focused on the modules I’ve just zipped through, allowing me to fast track to more meatier fare. Very clever. Once again, I’m seriously impressed. If kids are getting this from age 6 then it’s time to accept those of us significantly over twenty years old are dinosaurs-in-waiting!
I end up taking a break for a couple of weeks to work for a client (got to pay those bills). When re-entered the fold it became clear what had sunk in and what hadn’t. From the outset I’ve wanted to ensure I grasp the fundamentals of the subject matter and avoid rote learning.
In this case it’s primarily about understanding and manipulating quadratics. Realising their usefulness for calculating future outcomes in real world scenarios with multiple variables. For instance; physical trajectory when both gravity and propulsion are in play, and financial outcomes which are affected by many inputs (demand, margin, costs, etc).
The point of mastery is when I easily move between standard form and vertex form to calculate maximum/minimum values. This allows me to quickly pinpoint the optimal pricing point for a business to maximise their profits along the supply curve for instance. I’m not a theoretical mathematician so anchoring in the real world is gold dust.
Nail it. Move on.
I knew it wouldn’t be efficient to complete 100% of the algebra curriculum, I’m keen to truly grasp the fundamentals of a subject and then move on to the next step in my project plan. But the $64,000 dollar question is what percentage represents fundamental understanding?
I intended to apply Pareto’s Law as a rule of thumb, i.e. move on once I’d completed 80%. But when I got there I didn’t feel that could clearly see how all the modules fit together, partially yes, but my internal map was still cloudy. So I keep going.
It turns out that for my learning journey, 92% was the sweet spot. By this point I had mastered all the tools and understood how to move between them to solve different scenarios.
To help, I created my own data tools reference document that enabled me to quickly see which ways of manipulating quadratics would answer which questions (“What’s the growth rate within this period?” “How long will they take to reach maximum profitability?”). I even began drafting my own exercise questions in my head. I’ve done a lot of sports coaching and I found the best way to cement what students have learnt is have them teach others who are starting from scratch. Practice repetitions as a coach creates a fundamental understanding better than anything else.
So 92% was more than 80/20 but I didn’t mind taking more time at the beginning of my journey. Just like love you know when you’re there!
On to the next subject, basic computer engineering with Harvard’s CS50…