The 3 Levels of Data Science
In practically every industry in our current era, there are countless data points being documented and stored to improve workflow processes. Never before have we had so much opportunity to analyze human and machine behavior in order to optimize the way people experience products and services in their everyday lives.
The importance of the role data science plays cannot be understated; in fact, the field is growing so fast that McKinsey Global Institute (MGI) predicts that the US may face a 60% gap between the supply and demand of deep analytic talent in the near future. Some thought leaders have stated that the data scientist role is going to become the sexiest role of the 21st century.
Though the potential uses of data are varied, vast, and still expanding, there are three general tiers we can use to understand the different major functions:
Level One: Basic analysis
At the basic level, data science can be used to efficiently collect information around lingering questions your business currently has. For example, as a branding exercise, it can be used to gather information from social media sites to better understand how the outside world perceives your company.
Alternatively, it can gather feedback upon why customers flock towards a particular product, or avoid others completely. This type of analysis can help companies adjust on the fly based on the information gathered.
Level Two: Predictive analysis
Going more advanced, businesses can use data to predict how new changes made can potentially impact the performance of business elements in the future. Some of the questions predictive analysis can answer may include:
If price points are moved around, what is the likelihood of an improvement overall sales?
How can logistical bottlenecks be avoided in manufacturing plants by tracking the real-time functioning of vital equipment?
How will shipping costs change in response to rerouting due to an extreme weather event?
By answering these types of questions, organizations can be prepared to put out fires more quickly than they otherwise would have. However, there is another level of analysis that can go even further in terms of problem-solving capabilities.
Level Three: Proactive analysis
At this level, the machine learning is so advanced that it can begin to detect and facilitate the fixing of its own problems.
A great example of this would be: Suppose a jet engine is found to be running inefficiently, and burning excessive fuel unnecessarily. Through proactive analysis, the system could not only identify this problem, but also notify the proper mechanic to arrive at the airport where this plane will be landing next, and bring the proper equipment to alleviate the issue.
Through proactive usage of information to drive decisions, it can lower costs, prevent bumps in the road, and improve the overall user experience a customer has with your business.
Founder & CEO @ MachineMaze - Building the next generation in HiTech Manufacturing-B2B
3 年Tom, Interesting!
Artisan Agrarian
9 年I would suggest the first level be to consider the problems to be solved or the pressing questions to be answered. We currently have a data set that misses much of the real needs of our time. As with Florence Nightingale and patient care, the data set needs to be strategically established. What is the world we envision if we do fantastically well in progress? Lets empower those specificly.
BI Consultant
9 年Cool! For level 3, one should be a kind of thinker/researcher who essentially tries to understand all the data points, their interactive relationships and events that impact these data points, IMO. Such a person straddles both IT and business but will have BI skills. In real world, neither business nor the IT person will have aptitude/attitude to look at from both side and the role is not necessarily representative of a Bus. Analyst. Just my view from experience.
Industrial Digital Transformation & Manufacturing Data Enthusiast - Founder @Joltek, SolisPLC | Co-Host @ManufacturingHub
9 年Great post Tom. I've always enjoyed the combination of automation and data science. I had programmed a system which kept data of its own performance and made adjustments to the motor gains in order to compensate for the mechanical wear. This would greatly extend its life expectancy, but obviously there was a threshold at which the part would have to be replaced by a tech. Foreseeing issues and taking preventive steps is an art.