< PrevNext > Machine Learning & Predictive Analytics Will Help Travel Buyers Close Intelligence Gaps By Pi Ltd. director Al Norman / 27 January 2017 Share Corporate travel management is a data-intensive practice that has at times resembled an arms race between suppliers and buyers. Many travel managers have closed the data disparity that opened between them and suppliers several years ago by appointing analysts to their teams. Now, a new gap is opening as suppliers leverage their larger data sets and greater budgets to apply machine intelligence to analytics. It is about to become an imperative that travel managers do the same.Machine learning and predictive analytics are complementary but separate technologies. Machine learning is better suited to transforming or extending data sets to make them more useful, while predictive analysis allows businesses to improve the way they use these transformed data sets to deliver value.Suppliers are already using these advanced techniques daily, primarily to adjust the prices of their products each day or even each hour based on predicted and actual demand. Suppliers are also able to use the wealth of data from sources like loyalty accounts to shape traveler behavior.However, these same tools also can help the travel manager cut through the complexity inherent in today's complex programs. Areas where these techniques can help manage travel are:Building a total cost of trip view by traveler across multiple data sources and refining what constitutes a trip over timePredicting future activity based on trends or external market factors and telling business users when they could be impacted, thereby enabling them to prepare or even prevent the change, such as predicting flight delays through weather and past airline performanceIdentifying when suppliers' dynamic pricing models will result in advance-purchase opportunities, prices below negotiated or selloutsPredicting travel budgets based on performance, industry forecasts and future staffing levelsMatching individual transactions across agency, corporate card and expense data to eliminate duplication and clearly understand total spendPredicting the impact that events like the Olympics, the Super Bowl and corporate conferences will have on pricing and availabilityCleaning up data sets in which the data has been too 'dirty' to carry out meaningful analysisLeveraging data using advanced techniques will help mature programs find incremental value through lower cost, consistent quality and reduced risk. Driving these advanced analytics requires data. Companies can get started by scoping what's possible:Analyze available sources to find how to build links between themDetermine the data quality. Think about completeness of data rather than the presence of unstructured text. Machine learning often can work with the latterLook at what you might want to predict. Start with a long list and perhaps work with a business intelligence partner to refine what is feasible.