The Basics of Business Analytics - Part 2
By Chris Hedge
Techopedia defines data analytics as the “qualitative and quantitative techniques and processes used to enhance productivity and business gain”. Typically, data is collected and used to identify and analyze patterns via means ranging from simple visualization to sophisticated analytical models. Unfortunately, the term “analytics” has emerged as a rather generic term that vendors like to apply to products to differentiate their products or services from those of their competitors.
Business analytics can be broken down into four categories: descriptive, diagnostic, predictive and prescriptive. In the first of this two-part blog series, we focused on descriptive and diagnostic.
We'll now turn our attention to the final two: predictive and prescriptive.
Predictive Analytics builds on the diagnostic and uses that knowledge of causality to project what is likely to happen in the future given a set of assumptions around the causal factors. If we have determined that running a marketing campaign results in a 2% uptick in sales, then I can plan to run a marketing campaign in the next quarter and expect to get similar results in that next quarter. Granted that this is a simplistic view, but it is basic predictive analytics, and everything builds off this as much more sophisticated technique are applied.
Another example of predictive analytics is around equipment failures. Based upon descriptive and diagnostic analytics work, auto manufacturers know that certain parts of a vehicle fail after a certain amount of use. They can look at your mileage and maintenance history a determine with some degree of accuracy, what is likely to break next and approximately when. This manifest itself in your frustration when you take your vehicle in for an oil change and they tell you five more things you need to “fix” that aren’t yet broken.
Predictive analytics is not an exact science. There is a reason that you often see the Safe Harbor clause on the first page of a presentation containing any kind of forward looking statements, forecasts or predictions about future performance. Predictive analytics often uses past behavior to model how something may act in the future under those same circumstances but becomes much more difficult when trying to model behavior that has no historical base.
"One of the most disheartening parts of being in the forecasting business is that, despite all your hard work, your forecast will most likely be wronG."
I have done a lot of predictive analytics in the form of commodity supply, demand, and price forecasting in my career. One of the most disheartening parts of being in the forecasting business is, despite all your hard work, your forecast will most likely be wrong. Not because you’re not good at it, but rather because there is a plethora of overlapping and interrelating casual factors and it is extremely difficult to predict the occurrence of some of those (e.g. government policy, weather or natural disasters), particularly in combination.
Who could have predicted that two hurricanes hitting the U.S. Gulf Coast in 2005 would be the genesis of a U.S. energy resurgence?
Prescriptive analytics is the culmination of the advanced analytics efforts designed to answer the question -- “What should we do?” -- to take advantage of a future opportunity or to mitigate a future risk. It can involve an array of approaches from heuristics or optimization to decision and game theory, to machine learning and artificial intelligence.
Prescriptive analytics can be complex to administer and because of this most companies are not yet using them in their daily course of business. However, when correctly implemented these analytics techniques can have a very significant impact on how decisions are made and reflected on the company bottom line.
Prescriptive analytics can be used to optimize a value chain and recognize cross silo opportunities that you may not normally see. A good example is a commodity distribution system where you have a gasoline trading group and a diesel trading group. Each group has their dedicated logistics coordinator that manages the truck, rail, barge and pipe distribution of their respective products prescriptive analytics would drive them to optimize across the enterprise to find the most cost-effective way of meeting the delivery commitments.
In small companies with simple distribution systems this can be managed with spreadsheets and good communication. In larger companies with multiple commodity businesses managing distribution from multiple plants to multiple delivery terminals across multiple modes it can very quickly get complicated.
A good prescriptive analytics model would then allow you to plan multiple scenarios, quickly deal with disruptions, and understand the impact of asset acquisitions and divestitures. It would also Identify and better quantify risk associated with both short- and long-term decision making and assist in developing potential risk mitigation strategies.
But it is not all sugarplums and buttercups. Predictive Analytics can be very hard to implement, and the math involved can be beyond the capabilities of most mere mortals. Luckily, technology is a big enabler and toolsets are being constantly improved and simplified so that business people (not Data Scientist) can, once implemented, manage them with confidence.
Finding opportunities to apply analytics that can yield big dividends is not difficult once you know where to start looking. In the book, titled “Analytics at Work: Smarter Decisions, Better Results", by Thomas Davenport, Jeanne Harris and Robert Morison, the authors state that “you can find analytical opportunities by assessing your business decisions (regardless of their association with well-defined business processes) and asking how better information and analysis might yield better results”.
They suggest that opportunities can be found where there are:
1. Complex decisions with lots of variables and steps
2. Simple decisions in which consistency is either desirable or required by law (such as non-discriminatory credit and lending)
3. Places where you need to optimize the process or activity as a whole (especially when decomposing and optimizing locally cause you to sub-optimize the whole)
4. Decisions in which you need to understand connections, correlations and their significance
5. Places where you need better forecasts or anticipation
Irrespective of what business you are in or how technologically savvy your business is (or isn’t) chances are you are already using some form of business analytics to reduce costs, increase profitability and/or improve risk management. But, it is a dog-eat-dog world out there and to compete effectively, companies must further embrace the power of business analytics or risk getting mauled by a competitor.