DATA TITLE: Data underlying the research on Portfolio-based Airline Fleet Planning
DATE UPLOADED: 20 Jun 2019
AUTHORS: C.A.A. Sa, B.F.Santos and J-P. Clarke
The data is divided into four folders, according to the case study described in the MSc thesis "Robust fleet planning under stochastic demand" and in the OMEGA journal article "Portfolio-based airline fleet planning under stochastic demand". These folders are:
> Model 1 - Demand Forecast
> input_pax_per_route.xlsx - file with the raw data used to calibrate the demand model
(Raw Data) historical passenger data extracted from the TranStats database of the Bureau of Transportation Statistics (BTS/US DOT), which contains monthly scheduled US domestic passenger data based on a 10 percent ticket sale information dataset, aggregated for all airlines for the period 1990-2014.
(Routes_Model) consolidated demand for the 10 routes under analysis, using the data from ‘Raw Data’ sheet.
> estimated_model_parameters_per_ODpair.xlsx - file with the parameters of the demand model
(df_triple_loop_output) results from the calibration of the mean reverting Ornstein-Uhlenbeck process model parameters per market. These parameters were estimated based on the values from ‘input_pax_per_route.xlsx’ and were used to generate the results from Model 1.
> Model 2 - Fleet Assignment
> parameters.xlsx - this file has ‘constant’ parameters used in all optimisation runs of the LP model solved by Gurobi (www.gurobi.com)
(demand) input demand matrix
(distance) input distance matrix between airports
(yield) input yield matrix (revenue $ / revenue pax mile)
(aircraft) features of the 3 aircraft types considered
(portofolio) number of aircraft per fleet in the portfolio considered
> Model 3 - Scenario Generation
> aggregated_transition_probabilities.xlsx - file with the computed aggregated transition probabilities used in the discrete-time Markov Chain in the scenario generation model.
(y2y_2015--2016) transition probabilities between the 10 bins in 2015 and the 10 bins in 2016
(y2y_2016–2017) transition probabilities between the 10 bins in 2016 and the 10 bins in 2017
(…)
> Yield Analysis
> Dataset_fares_2011_2014_Table1a_new.xlsx - BTS/US DOT fare data from 2011-2014. This data was used to estimate the yield for the routes considered. The division of fares by distance, results in the yield values analysed.
(raw data) historical data (from 2011 - 2014) of average fares for multiple OD pairs in the US market
(distance) distance, in miles, between OD pairs.
All this data was generated by the authors or extracted from public sources. The data can be distributed, remixed, tweaked, and built upon our work, even commercially, as long as you credit the original work - OMEGA journal article "Portfolio-based airline fleet planning under stochastic demand.
For information, contact B.F.Santos (b.f.santos@tudelft.nl)