By Jamie Khoo | August 1, 2019
There are too many articles out there emphasizing the need for employees to ‘evolve or be automated’ and too few guiding how employers & managers should help their employees evolve. This stems from both employees & employers/managers not being fully informed of the true capabilities of automation and data science. It’s interesting because some claim the data science movement itself is losing steam – a.k.a. data science is not the right future for data analysts. Hence, the first question to crack is: how does the future of data analysis look like?
To answer this, one must first understand the driving forces behind automation and the ‘boom’ of data science. Looking back, the discipline of data science has always been present; the idea of using software to assist with iterative calculations or processes is not new (i.e. SPSS, R, Python itself is 20 years old). How do we explain this - what caused the sudden boom in data science and quick automation? In my opinion, three factors drove this ‘boom’:
Age of information: The Internet enabled a stream of information that encouraged self-education. Through the internet, one is able to learn how to code, pick up statistical concepts, and understand machine learning concepts to a certain level.
Moore’s Law on hardware acceleration: Moore’s law dictates capacities of computers double every two years. Given this law, consumer grade computers can handle more demanding processes & calculations. With this acceleration, consumers can now tinker with scrappy homemade (read: barely optimized) algorithms without paying for expensive hardware simply because consumer-grade hardware is powerful enough.
- Upgraded formal education: This is a longer term macro-level trend. Formal education has a solid base of mathematics and with the 1990s & 2000s increase of college graduates in the workforce, we can observe mandated college level mathematic courses further boosted this ‘base’. With that, we have an entire workforce ready to understand the concept of multiple dimensions that Machine Learning brings to the table. In short, if everyone understands Cartesian space, grasping the concept of multidimensions via Euclidean spaces isn’t too much of a leap
- Formal education is about to be upgraded again with the idea of code being taught to children in schools. With this, math AND code will form the new ‘base’ of knowledge for the future workforce.
Knowing these factors, the answer for managers looking to adapt their team of analysts for a future of automation and data science lies in these three reversed-engineered points:
Upgrade your team’s formal education – keep up with the times. Thankfully formal education these days does not require us to leave our full-time jobs. There are night courses under ‘professional studies’ to assist with pursuing formal knowledge without sacrificing one’s job.
Age of information – use the internet to hack your way in. Once you have the ‘base’ of formal knowledge, supplement that with things you can learn on your own. Two examples: complete side projects and compete in Kaggle contests. A data scientist’s true power lies in their ability to apply knowledge. For example, KPMG places a higher value on applied knowledge vs formal knowledge. Internet doesn’t need to be the only source here - if you have a large enough team, create a culture of knowledge sharing. Introduce specialization and cross-training.
Upgrade your hardware. This is probably the easiest and most obvious – but if you want to attract talent who can will the latest technology for your purposes, you’ll need to first supply them with the latest technology. It is a straight-forward formula: study what caused the ‘automation and data science’ wave, then reverse engineer them to ride the wave.
Going back to the first question: what does the future of data analysis look like? The future has always been the same: ability to apply the (latest) knowledge. The spoils have and always will go to those who can fuse formal knowledge with real-world needs, and can communicate these concepts in a simple, succinct format. Presently, ‘knowledge’ just happen to also include ability to code and apply machine learning concepts.
Written by Jamie Khoo, Director of Marketing Sciences, PHD. Opinions are his own, not representative of Culture of Insight or PHD (OMG) Media