Applications of various Time Series Models in the context of prices of airlines ticket like low budget carrier.
Time plays an important factor when we consider of prediction in the field of Aviation. All the Models other than Time Series Forecasting do not deal with time. A seat selling for any price above the service cost per single passenger would be profitable for the airline
Time Series is a non-causal model and a stochastic process where time is an important independent factor. This is widely used across the industries for demand forecasting, pricing strategy etc.
The dynamic pricing is based on many of following factors such as:
- Direct Competition
- Departure schedules (Day, time,month etc.)
- Seasonality ( Festivals/ holidays)
- Type of route (Direct/ multiple stops) etc.
- Oh Yes Now God’s Impact Pandemic aka Covid19
Time series model plays an important role in determining following for Aviation sector:
- No. of airpassengers (Demand forcasting)
- Dynamic air fares
- Airline delay
Few Examples
Over the past 20 years the low-cost airline model has grown excessively. Among the successful low-cost carriers Southwest Airlines, which operates within the us and Ryanair, which operates in Europe are often named. The LLC model addresses to business strategies, which lessen the value of the airline. Typical cost-saving practices include :
#Operating at secondary airports;
#Flying a single airplane type;
#Increasing airplane utilization;
#Relying on direct sales;
#Offering a single-class product;
#Avoiding frequent-flyer programs;
#Keeping labor costs low.
Indigo A case study
Airlines like Indigo, allow booking tickets in various categories which differ not only in their price but also booking conditions and privileges. A Passenger buying higher price ticket is more profitable to an airline however, these passengers usually book their tickets in the very last minute and are very few compared to the economy class passengers. It is therefore important for Airlines to balance price with demand. In this regard it is important to determine
- How many seats be reserved for each booking class so as to minimize the risk of rejecting a low class booking for a possible higher class booking. This directly impacts the number of booking per category allowed by the airline.
- Forecasting demand for No-Show and Cancellations scenarios. This helps the airline determine the optimal overbooking that can be allowed thereby creating a balance between unoccupied seats and denied boarding.
The forecasting demand in the above two scenarios engages statistical modelling techniques like Time Series forecasting which considers historically data of similar flights. However, a large number of time series forecasting methods are available, some of which are detailed below:
If we don’t limit our problem statement for just price determination, we can build forecasting model for engine maintenance by the means of which we can schedule/alter the flight route for aircraft, discounts that can be given without hampering our profits and the seat allocation system i.e., to find which seat is to be kept for what price during web check-in for any given season/day to fetch maximum benefits.
Conclusion
In real life scenario, airlines pricing are very unpredictable and keeps on fluctuating. These fluctuations could be because of any external factors which are beyond the control of the airlines. Due to these factors there will always be irregularities in the prices. To forecast the pricing with irregularities ARIMA model is used, in this model using the ARIMA (p,d,q) where p is number of partial correlations, d is number of differencing done on the series and q is the number of significant correlations the forecasting is done. Using the ARIMA model, we can forecast airline prices for a short period of time and should avoid forecasting too far into the future as here the model can give invalid results.
