1 Home

Introduction

Imagine an airline company that wants to decide on the airplane capacities of different routes for the next 12 months. The decision to make now is how many airplanes of each type, the company is going to contract in the fleet for the next year. The actual assignment of the airplanes to routes can be decided later on a monthly basis.

Demand Prediction using ARIMA Model

As a first step to optimizing the airplane capacity for the next 12 month, a linear time series model should be fitted to predict the passenger volume.

We can see that the variance of passengers number is increasing, so we can transform the time series using log transformation to make the process more stationary.

Model the process using auto.arima

tsPassVol.log <- ts(datf$passVol.log, frequency = 12)
modARIMA <- auto.arima(tsPassVol.log)

Scenario Generation and Reduction

Deterministic prediction can reflect the information about expectation. To use variance for prediction, scenario generation and reduction has to be used.

Stochastic and Robust Optimization

Stochastic Optimization

Robust Optimization