PLF (Passenger load factor) measures the capacity utilization for airlines. It indicates the efficiency of the airline; filling seats and generating revenues. 80% of passenger load factor is considered as standard in the domestic airline industry.
The improvement of PLF is a direct impact on the bottomline. The bettering of this KPI effects the plan and costs of complimentary functions; such as workforce, fuel, catering and ground services.
PLF is highly effected by seasonality, unpredictable demand and even political & economical issues. Considering the airline industry dynamics as cancellations of reservations, multi-leg flights complexity and the openings of new flight routes, forecasting PLF becomes even harder. Yet it is attainable by building airline-specific models to forecast the aggregate passenger traffic in a certain time frame, region or an individual flight. The model delivers an optimum revenue and enables the business units to cleverly act on price and campaigns.
The historical data and the current reservations of the 1500 flights are supplied to the model everyday, releasing a 365 day window of forecast. A hybrid adaptive regression model is built to forecast PLF with the smallest error margin possible. The model is delivering an 80+% PLF – above the industry standards - and there is a 20% reduction in the forecast errors. 59% of the flight based forecasts are reported to be within the 0-10% error slot.
As the forecasted period is 12 months ahead airlines have enough time to optimize the number and the type of aircrafts appointed to the routes. Determining the optimal number of flights helps in better scheduling the flights and effective capacity utilization.
With PLF Optimization, a load factor prediction model has been developed in order to better manage the flight capacity. Since a dynamic and conditional model method is chosen, the daily job of the models is trained by parallel processing tools.
The model is developed on R and the visualizations are developed with Microstrategy.