Forecasting of Airline Passengers Based on Machine Learning

Abstract

The management of airlines depends on the forecasting of air passenger flow, but standard forecasting techniques cannot guarantee the accuracy of the forecast. When they encounter large-scale, multidimensional, nonlinear, and non-normal distributing time series data, they have the ability to generalize. In this paper the SVM regression is implemented to help with forecasting air passenger flow. We discover that the SVM regression algorithm's outcome exhibits the least inaccuracy when compared to the other two forecasting techniques by carefully choosing the parameters and kernel function. Concerns about demand splits among service providers have been given to every sector by the nearperfect competition scenario. This is especially important in the airline business, where high service standards are the norm. Demand is driven by the player who achieves the greatest mappings between every one of his the airline's offerings and the set of consumer preferences. The airline business has grown exponentially as a result of the economic reforms of 1991, which were promptly followed by the privatization of Indian skies, creating nearly ideal competition. More specifically, cross-border activities have started in the international sectors, where previously only domestic carriers operated. Keywords – Airline Passengers, Support Vector Machine, Forecasting, Machine Learning