Once you have got your descriptive and predictive analytics together, you have the information you need to be able to start making business decisions. However, it is still very easy to overlook an option or make less than optimal decisions. It is in the pursuit of making correct choices that prescriptive analytics become vital.
Prescriptive analytics systems take advantage of descriptive and predictive information to suggest decisions which capitalise on the predictions. This means difficult decisions can be handled much more easily, taking into account many more factors than a person could handle, and without human bias.
What does it look like?
Prescriptive analytics is a relatively new field and good examples of it can be few and far between, hence the competitive advantage well implemented prescriptive systems can provide. One nice example of prescriptive analytics can be seen in the journey planner apps that advise on transit directions. These consider the multitude of different ways you could get from A to B by bus, tube or on foot, first predicting how long each method would take, then suggesting the best route based on time taken, cost or even how long you’d be outside for if it’s raining.
“#Prescriptiveanalytics is the essence of data driven decision making.”
This is the essence of data driven decision making and integrating these kinds of systems can improve business performance in a range of areas: from how to most efficiently deliver your product to your customers to how much raw material to buy to meet demand. It is the ability of these systems to make important decisions like this, based on real evidence, quickly and accurately that has ignited the excitement about data science and created the coveted job of “data scientist”.
Prescriptive analytics use a lot of the same techniques as descriptive and predictive analytics to form their prediction, so there is a lot of overlap in terminology. A couple of the more specific terms are shown below.
|Optimised||A solution which produces the best possible outcome. For example, the route that takes 5 minutes less than any other.|
|A/B Test||An experiment often used to inform prescriptive systems. Show one group of customers your new website while keeping the old one for another group, you can then use the results to inform the decisions made by a system.|
|Exhaustive search||A basic prescriptive technique where every possible solution is analysed in order to find the best one. For example, calculating the length of a journey via every possible combination of trains, cars, buses etc. and suggesting the shortest one. Very inefficient but guarantees the optimum result.|