Do You Want to Know the Future?

STOCKHOLM - Dec. 12, 2017 - PRLog -- Remember the movie "Back to the Future"? In the second episode, Biff got the sports almanac and used it to bet on the games knowing their results. Would it help your business, if you knew how many clients are you going to have today, how much short life products are you going to sell in three days, what the call for these new apartments would be? I bet it would.

Machine learning and Big Data analytics are gradually changing the landscape of many businesses. This avalanche is just beginning its path, but we can already be blown away by its power.

According to the new research from GE and Accenture*:

• 87% of companies believe Big Data analytics will redefine the competitive landscape of their industries within the next three years.
• 89% of enterprises believe that companies that do not adopt a Big Data analytics strategy in the next year risk falling back and losing their market share.
• Increasing profitability (60%), gaining a competitive advantage (57%), and improving environmental safety and emissions compliance (55%) are the three highest industry priorities in implementing Big Data and Machine Learning initiatives.

* - Source: "How the Industrial Internet is Changing the Competitive Landscape of Industries".

Many companies have already started looking into Big Data analytics to find out what benefits is can bring them. For example for aviation, Big Data in a conjunction with Machine Learning opens endless capabilities for improvement. It can help with trends detection. Applying it to aircraft fuel consumption data will give airlines the ability to see the fuel burn increase even before it has passed a critical threshold of the average fluctuation. Passenger booking data will give an insight if the sales are going well and the flights will have the highest load factor possible. Moreover, it opens a way of efficient overbooking management and allows engaging customers with personalized shopping experience and dynamic offer management. Not to mention extraordinary savings of the predictive maintenance approach.

Airline industry is just one example sector that can put Big Data to good use. Potential use cases and implementation areas are endless, including retail, logistics, transportation, procurement, and more. We decided to be proactive and came out with our own solution that would demonstrate actual improvements from Big Data analytics. The solution we developed is advantageous for retailers, caterers, airlines, and other passenger transportation companies. It helps with planning the procurement of perishable products and thus allows avoiding losses due to products shelf life expiration or short quantity.

Why planning the procurement of perishable products? Existing procurement planning has accuracy of about 70%. With this level of accuracy, retailers, caterers, and others dealing with perishable products may buy excessive or insufficient amount of certain products. In such cases, they suffer a loss when product self-life expires or miss profit when they are out of stock. Increasing planning accuracy would result in savings on costs, enhanced customer satisfaction, avoidance of loss of profit, uninterrupted continuity, and more.

We began with a hypothesis that sales of products of different categories vary and should be described with different mathematical models, and mathematical model predicting sales of a particular product category can change irregularly. So, we have built the solution as an adaptive composition of simple adaptive models. In other words, several different mathematical prediction models were matched to predict sales. The solution is adapted to use the model that demonstrated the highest accuracy over the last week to make predictions for the week to come. To avoid overfitting, an optimum amount of models is evaluated with the Combinatorial Theory of Overfitting.

The developed sales prediction solution (https://sigma.software/case-studies/procurement-planning-...) was tested on real-life data and proved successful. Sales prediction accuracy improved by more than 10%. As a result of improved accuracy, the loss caused by incorrect procurement decreased. Moreover, since the solution used simple mathematical models for predictions, requirements for hardware resources have also decreased. The choice of simple mathematical models was conditioned by our striving to make the most efficient solution for clients, minimize development time and cost. Other possible ways of implementation, such as neural networks and deep learning would skyrocket the cost and require incomparably more powerful hardware resources.

This is just one example how Machine Learning and, specifically, Big Data analytics can improve business and financial performance. Believe it or not, but every 5th task performed in IT systems can be automated or improved using Machine Learning or Big Data.