Predictive Analytics, what can it do in the real world?

Introduction

Cognitive bias is everywhere, especially when in comes to big decision making and even more so when large amounts of data are available. Without the ability to properly crunch numbers into useful answers, it’s easy to bring in bias or keep doing the wrong thing. 

How poorly our brains are at data comprehension is nicely illustrated by the question “how far, if it was possible, would a stack of paper reach if it were folded 50 times?” It’s possible to imagine, even if told that at 10 folds, the paper were the width of a hand. Our brains are not possible to compute the geometric progression that makes the stack of paper at 50 folds actually reaching the sun, 51 folds would reach from the sun and back. 

With more and more data being generated, these types of cognitive fallacy make prediction of data sets frustratingly impossible without assistance. 

Predictive analytics is aimed at not only processing data more accurately, but using that data to build models that can more accurately simulate. 

Predictive analytics can be used in many different ways to enhance efficiency, which, in the future, may have an impact on how things are sold and consumed. 

Industries

 “Near Zero downtime” in Manufacturing 

FANUC is a Japanese automation company, focusing in manufacturing, in 2019 they released their product of zero hours downtime. Using operational data, predictive analytics is used to provide lists of potential problems to clients. Customers share their data, which is processed in the cloud to then run though the predictive algorithms. 

One of the main customers of this service was General motors who connected one in four of all its 300,000 factory manufacturing robots to this zero hours prediction engine. Since then it has attributed 100 potential failures that were prevented to the new system in 2 years. 

Alongside that, the main cost savings were from GM being more JIT (just in time) as stock movements and demands were calculated more accurately, storage costs were also reduced.

Transporting 15% of GDP 

Maersk Line is an international shipping container with more than 600 vessels, this accounts for 15% of global GDP that they transport. 

They rely heavily on data analytics to find efficiency due to their large cost base. 

They use predictive analytics to reduce the operational costs by increased efficiency. They use AI to analyse the 2 million containers they use and to calculate the optimal way to deliver to 350 ports around the world. The algorithms are aimed at calculating fuel economy, voyage optimisation, reefer containers and empty container optimisation. 

This data centric approach as lead to $100 million cost saving, with the company hoping to expand the technology in the future. 

Nike and their ‘Direct Offence’

Whist the number of uses of predictive analytics for nike are countless, as their whole supply chain would be data rich, they recently focused their supply chain, leveraging AI tools such as predictive analytics to gain more charge of their customer relationships. 

They are using anticipatory shipping in future, where they ship the items before the customer orders them because they can predict purchase behaviour with such accuracy. As the VP of digital products of Nike says: 

“We’re borrowing from the digital to do in the physical.”

— Michael Martin

These benefits are statistical and proven in probability, saving and generating large amounts. 

Implementation

The below diagram shows partners model of analytics and how they transform over value.

The previous examples show examples of the whole range of ascendancy, from hindsight to foresight. 

The important realisation with this graph is how it shows a trajectory of data usage, by building an information culture along with systems and infrastructure, the following objectives can be delivered, as leading into diagnostics, into prediction. 

Examples

By naming a few examples of where Predictive analytics can be applied (to name a few) it can outline how simple collection of early data can be transferred into powerful analytics. 

  • Inventory optimisation 
  • Fraud detection 
  • Customer lifetime value identification 
  • Product recommendations 
  • Customer retention 
  • Demand forecasting 
  • Credit applications 
  • and much more…

 With the amount of data being generated increasing all the time, the amount of company data that can be empowered will increase. 

Predictive analytics will become more adopted as comprehension and accessibility become greater. 

Tips for success

Since tangible benefits are being felt in many industries the main objective to start adoption is to start at the beginning. Data collection is either through using existing data or generating new data streams to supplement the analysis. 

Startups are becoming more used to involving systems to mange low levels of data, with the knowledge that as the users or sales increase, so will the data streams, having a pipeline or infrastructure in place can create a foundation for efficacy and cost generating systems to sit within. 

To see how a smaller sized project can be the best option to move forward, read our article on MVP design and the top reasons for success using a MVP designed process click HERE

Otherwise, for more information on predictive analytics and how it can impact a company or industry please email info@proai.co.uk to find out more. 

Leave a Reply