Planeación óptima de la red de rutas de una aerolínea
An analytic methodology is developed to recommend the optimal plan for an airline network, determining the best capacity deployment by route and by month in one year horizon, having the objective of maximizing the projected profitability. This Work comprehends the integration of several predictive m...
- Autores:
-
Rojas Arcila, Daniel Camilo
- Tipo de recurso:
- Fecha de publicación:
- 2020
- Institución:
- Universidad de los Andes
- Repositorio:
- Séneca: repositorio Uniandes
- Idioma:
- spa
- OAI Identifier:
- oai:repositorio.uniandes.edu.co:1992/51034
- Acceso en línea:
- http://hdl.handle.net/1992/51034
- Palabra clave:
- Líneas aéreas - Planificación
Transporte de pasajeros - Colombia
Aviación comercial
Ingeniería
- Rights
- openAccess
- License
- https://repositorio.uniandes.edu.co/static/pdf/aceptacion_uso_es.pdf
Summary: | An analytic methodology is developed to recommend the optimal plan for an airline network, determining the best capacity deployment by route and by month in one year horizon, having the objective of maximizing the projected profitability. This Work comprehends the integration of several predictive models with a prescriptive model of Mixed Integer Programming (MIP), which at the end will provide the optimal recommendation for the capacity deployment of the airline. The projected profit is modeled in the MIP as the criteria to be maximized, however, to model the revenues, it is necessary to answer three fundamental questions: 1. ¿what will be the capacity deployment of competitors in each route for the following year? 2. ¿How is the passenger?s response versus market capacity changes? 3. ¿What has been the historical effects that capacity changes have in average fares for each route? To respond to these questions predictive models are developed; for the first question, four models involving time series and neural network are applied for each route (Auto-Arima, Prophet, LSTM, MLSTM), every model is assessed by the Root Mean Square Error (RMSE), assigning the model with the best performance in each route. On the other hand, regression models are implemented to model the passengers and fare in terms of capacity. Once the models for competitors offer prediction are obtained (question one), a potential range of several choices of capacity deployment for the airline are combined with the capacity projection of competitors to obtain potential total market deployment, which is used to estimate passengers and fares by using the regression models. Finally, these models provide the estimations to the MIP and then the optimization model is run to provide the plan recommendation. |
---|