Predictive analytics: crystal ball or reliable planning tool?

More and more data along value chains is being gathered and analyzed, chiefly in industry and commerce. Companies collect their own data and increasingly using data from upstream and downstream links in the value chain, including logistics. Data can now be used to simulate potential future scenarios. Does that sound like looking into a crystal ball? It’s actually pretty high-tech.

The amount of data being generated is steadily rising, and IT is providing ever more ways to analyze it so it can be put to use. What was known as “business intelligence” in the 1990s is developing into business analytics, which is increasingly enabling predictions about future system statuses. Business analytics can use new data and data analysis technologies to make predictions about future likelihoods – in other words, becoming predictive analytics.

It may sound complicated and abstract, but we’re already familiar with and rely on one application of predictive analytics: the weather report. Weather forecasts have improved considerably in recent decades. On the one hand, this is because more data can be collected, for example using powerful weather satellites. On the other hand, computer technology has become more powerful so that the vast amount of weather data can be used to calculate accurate forecasts. That’s all that predictive analytics is – it calculates and simulates future scenarios and their likelihood.

Thomas Reppahn, Head of Logistics Product and Process Management, Schenker Deutschland AG

“Predictive analytics has revolutionized the potential of business processes, particularly in logistics.”

More data means a better forecast

Predictive analytics depends on the volume and reliability of raw data. There is still much more raw data that could be collected, especially in logistics. Traditional data, such as arrival time, length of processing and load volume, is provided during transshipment. Logistics companies can also use their own company data from warehouses or warehouse management systems, which are often already automated or partially automated, and from their own plans, such as shift schedules, route plans and loading dock planning.

Logistik aktuell spoke with Thomas Reppahn, Head of Logistics Product and Process Management at Schenker Deutschland AG, about the decision support system already being used in contract logistics. According to Reppahn, “Predictive analytics has revolutionized the potential of business processes, particularly in logistics. Once data gathered along the entire supply chain can be used for model calculations, this will give everyone involved much more planning certainty. But we still have a long way to go before that happens. We’re already using these types of planning tools at our company. Our decision support system is just part of predictive analytics. It helps us make decisions by evaluating the data we already generate. It is already serving us well when it comes to our own planning.”

But much more is possible, namely using sensors, for example to monitor cargo. Sensor technology is becoming less and less expensive, which means that companies can increasingly afford to use it on a large scale in logistics. It can supply data directly to the logistics company’s computer system, in real time. Sensors could record considerably more data, which would be important for the entire supply chain, not just the logistics company. Retailers and producers alike would be interested in real-time data about the condition of cargo, such as temperature, vibration and humidity.

It’s not for nothing that a study on trends and strategies in 2017 by BVL International listed the following three technologies in logistics as having the most promising future:

  1. Predictive analytics
  2. Mobile data access for customers
  3. Monitoring sensors

Using predictive analytics to combat the dreaded bullwhip effect

The combination of sensors and predictive analytics in particular has a promising future. Together, these two technologies can be used to prepare accurate forecasts and simulate different scenarios that would be hard to imagine today – and for all participants in the supply chain.

“#FutureTrend in #logistics: #AI and #sensors for #PredictiveAnalytics.“

Tweet WhatsApp

At the very least, predictive analytics can be used to help eliminate the psychological reasons why the bullwhip effect occurs. These reasons usually occur when individual participants in the supply chain do not have the full picture. The more accurately we can predict future scenarios, the fewer precautions each of the participants in the supply chain has to take. This would eliminate some of the incentive for panic ordering.

Returning to our weather forecast comparison, weather reports used to be synonymous with miscalculations. Forecasts have become increasingly reliable over the years thanks to cutting-edge weather data and more robust calculations. Today we trust weather forecasts without a second thought. Likewise, predictive analytics can make it possible to predict normal business. The computing power and sensors already exist; now we just need to use them.

The bullwhip effect

The bullwhip effect is a supply chain phenomenon that occurs when minor fluctuations in demand at the end of the supply chain lead to greater ordering fluctuations in the upstream supply chain. These fluctuations build up as they move toward the beginning of a supply chain, similar to the motion of a whip. There are four reasons why the bullwhip effect occurs:

1. Processing of demand signals
2. Price fluctuations between individual links in the supply chain
3. Order bundling to reduce the fixed cost of ordering
4. Panic ordering due to a fear of shortages