About me and my work

Welcome to the my page. My name is Marek Tomas and I am a doctoral student at Faculty of Aeronautics of Technical university of Kosice in Slovakia. My passion for aviation has been going on since I was little and flying was always fascinated for me. I have never forgotten the feeling when I flew on airplane first time. It was about 30 years ago and I flew on Tunisair Airbus A320. I have never forgotten the moment when the pilot came out of the cockpit and greeted each passenger personally. I fulfilled my dream at the Faculty of Aeronautics and I continue to live it. Aviation is simply my hobby and my passion as well.

On this page you can find the some results of my doctoral work – the airport prediction model database. This database was created based on a questionnaire filled out anonymously by airports from all over the world. The questionnaire was part of the dissertation research at the Faculty of Aeronautics of Technical University in Košice in Slovakia.

The site is intended for airports, which can give a new perspective on the choice of a prediction method. The airports in the database are divided into 3 sizes based on the number of passengers handled and the number of flights. Selection by preference will result in a recommended prediction method used by similar airports.

Airport prediction methods database

Airport prediction methods
Filter
Field of prediction interest
Size of airports by carried passengers
# Prediction method Country of airport Passengers carried by airport
1 Time series analysis decomposition methods Austria 13 327 000
2 Delphi method Austria 20 000 000
3 Mixed integer linear programming model Brazil 13 500 000
4 Causal method Brazil 34 400 000
5 Time series regression analysis Brazil 34 400 000
6 ARIMA, SARIMA model Croatia 1 102 000
7 ARIMA, SARIMA model Cyprus 6 037 000
8 Time series analysis decomposition methods Czech republic 10 734 880
9 ARIMA, SARIMA model Denmark 3 700 000
10 Airhart, Turnaround optimization Denmark 22 063 000
11 Petri net model Denmark 3 700 000
12 Statfor - Statistics and forecast metohods Egypt 7 200 000
13 ARIMA, SARIMA model El Salvador 3 600 000
14 Time series analysis decomposition methods France 29 184 000
15 Time series analysis decomposition methods France 5 800 000
16 Time series analysis trend projection France 8 532 000
17 ARIMA, SARIMA model France 9 264 000
18 Time series analysis decomposition methods France 5 800 000
19 Time series analysis decomposition methods France 8 532 000
20 Time series analysis decomposition methods Germany 2 500 000
21 Time series analysis decomposition methods Germany 8 300 000
22 Time series analysis trend projection Germany 16 050 000
23 Time series analysis decomposition methods Germany 2 500 000
24 GVAR model Germany 11 080 000
25 Amorph Germany 11 000 000
26 Time series analysis decomposition methods Germany 63 000
27 Time series analysis decomposition methods Greece 2 780 000
28 Time series analysis decomposition methods Greece 2 791 000
29 Time series analysis decomposition methods Greece 3 749 000
30 Time series analysis decomposition methods Greece 2 780 000
31 Time series analysis decomposition methods Hungary 16 200 000
32 Time series regression analysis Hungary 16 200 000
33 Delphi method Iceland 6 126 000
34 Time series analysis decomposition methods Italy 3 077 000
35 Grey Markov Model Moldova 2 308 000
36 Time series analysis trend projection Nigeria 16 172 625
37 ARIMA, SARIMA model Philippine 33 757 000
38 Time series analysis trend projection Philippines 50 235 000
39 Time series analysis decomposition methods Poland 2 876 000
40 Time series analysis decomposition methods Poland 2 252 000
41 ARIMA, SARIMA model Portugal 28 000 000
42 Time series analysis decomposition methods Romania 2 644 000
43 ARIMA, SARIMA model Romania 620 000
44 Causal method Romania 2 644 000
45 SWOT analysis Russia 18 100 000
46 Time series analysis autocorrelation method Russia 24 000 000
47 Autocorrelation method Russia 24 000 000
48 ARIMA, SARIMA model Saudi Arabia 42 700 000
49 Time series analysis decomposition methods South Africa 21 000 000
50 Time series analysis decomposition methods South Africa 21 000 000
51 Causal method South Africa 7 157 000
52 ARIMA, SARIMA model Spain 60 220 000
53 Time series analysis decomposition methods Tunis 4 000 000
54 Time series analysis decomposition methods Tunis 4 000 000
55 Time series analysis decomposition methods Turkey 11 914 000
56 PLS-SEM path modeling United Arab Emirates 15 540 000
57 Time series analysis decomposition methods United Kingdom 800 000
58 Time series analysis decomposition methods United Kingdom 382 000
59 Time series analysis decomposition methods Uruguay 25 078 000
60 Time series analysis trend projection USA 2 400 000
61 ARIMA, SARIMA model USA 73 362 000
62 ARIMA, SARIMA model USA 12 398 000
63 Time series analysis decomposition methods USA 2 400 000
64 Causal method Vietnam 15 135 000
65 Time series regression analysis Vietnam 15 135 000

Description of airport prediction methods

ICAO methods

Decomposition methods

Decomposition methods involve the dissection of the problem into various components. These methods are particularly relevant when strong seasonality or cyclical patterns exist in the historical data. These methods can be used to identify three aspects of the underlying pattern of the data: the trend factor, the seasonal factor and any cyclical factor that may exist.

The time-series analysis methods

The time-series analysis methods are largely based on the assumption that historical patterns will continue, and they rely heavily on the availability of historical data. A first step when forecasting air traffic activity is usually to study the historical data (time series) and determine the trend in traffic development. In the context of medium-term or long-term forecasting, a traffic trend represents the development in traffic over many years, isolated from short-term fluctuations in traffic levels. When deriving a medium-term or long-term forecast by extrapolating from the traffic trend, the forecaster assumes that the factors which determined the historical development of the traffic will continue to operate in the future as in the past, except that their impact may change gradually, and steady-state conditions will continue into the future.

ARIMA

ARIMA, standing for Autoregressive Integrated Moving Average, is a versatile model for analyzing and forecasting time series data. It decomposes the data into three key components:
1. Autoregression (AR) - this component captures the influence of a series’ past values on its future values. In simpler terms, AR considers how past observations (lags) affect the current value. It’s denoted as AR(p), where ‘p’ represents the number of lagged observations included in the model.
2. Differencing (I) - stationarity is a crucial assumption for many time series analyses. Differencing involves subtracting a previous value from the current value, often required to achieve stationarity.
3. Moving Average (MA) - this component accounts for the effect of past forecast errors (residuals) on the current prediction. It considers the average of past errors (lags) to improve the forecast accuracy. MA is denoted by MA(q), where ‘q’ represents the number of lagged errors incorporated in the model.

SARIMA

SARIMA (Seasonal ARIMA) builds upon ARIMA’s strengths by incorporating an additional dimension: seasonality. This is particularly beneficial for data exhibiting recurring patterns at fixed intervals, such as monthly sales data with holiday spikes. Here’s how SARIMA tackles seasonality:
1. Seasonal Autoregression (SAR) - similar to AR, SAR considers the influence of past seasonal values on the current value. It captures the impact of past seasonal patterns on future forecasts.
2. Seasonal Differencing (SI) - analogous to differencing, seasonal differencing focuses on removing seasonal patterns from the data to achieve stationarity.
3. Seasonal Moving Average (SMA) - this component incorporates the influence of past seasonal forecast errors into the current prediction, similar to the moving average component in ARIMA.

Casual methods

Casual methods extensive use has been made of trend forecasting by basing judgement on past growth trends, which the analyst simply extrapolates, based on the historical values. In the short term, this approach appears to be reliable, especially when the extrapolation procedure is applied with modified growth rates to account for short-term disturbance in underlying trends. In the long term, this type of extrapolation is likely to be unreliable and is theoretically difficult to substantiate. Consequently, forecasts derived by taking into account how economic, social and operational conditions affect the development of traffic offer an alternative to time-series analysis.

The Delphi technique

The Delphi technique is a special procedure for forecasting by consolidation of opinions on the future. It has two steps. A selected group of qualified people are first presented with a questionnaire in which they are requested to indicate a most probable course of development in the activity being forecast. The initial returns are then consolidated and the composite response returned to all contributors giving them the opportunity to revise their original assessments in light of prevailing opinions among other experts. The Delphi technique is a practical means of bringing together information from many experts and moving towards a consensus among them.

Autocorrelation method

Autocorrelation method, sometimes known as serial correlation in the discrete time case, is the correlation of a signal with a delayed copy of itself as a function of delay. Informally, it is the similarity between observations of a random variable as a function of the time lag between them. The analysis of autocorrelation is a mathematical tool for finding repeating patterns, such as the presence of a periodic signal obscured by noise, or identifying the missing fundamental frequency in a signal implied by its harmonic frequencies. It is often used in signal processing for analyzing functions or series of values, such as time domain signals.

Non ICAO methods

A Grey Markov Model (GMM) is a hybrid model that combines the Grey System and Markov Chain theories. The Grey System theory is used to predict trend values, while the Markov Chain theory is used to forecast fluctuation values. This combination allows the model to provide forecast results that involve two aspects of information.The Grey prediction model, often expressed as GM, where u is the order of the differential equation and time series data.

The Global Vector Autoregressive (GVAR) approach, provides a relatively simple yet effective way of modelling interactions in a complex high-dimensional system such as the global economy. Although GVAR is not the first large global macroeconomic model of the world economy, its methodological contributions lay in dealing with the curse of dimensionality.

AMORPH utilizes a new Bayesian statistical approach to interpreting X-ray diffraction results of samples with both crystalline and amorphous components. The program simulates background patterns previously applied manually, providing reproducible results, and significantly reducing inter- and intra-user biases.

AIRHART is a modular total airport management platform that enables airports and all stakeholders to become smarter – together. AIRHART facilitates an effective eco-system for optimizing operations, passenger experience, commercial excellence and the ability to adapt to new needs – faster.

PLS-PM a co mponent-based estimation approach that differs from the covariance-based structural equation modeling. The measurement models represent the relationships between the observed data and the latent variables. The structural model represents the relationships between the latent variables.

A Petri net is one of several mathematical modeling languages for the description of distributed systems. It is a class of discrete event dynamic system. A Petri net is a directed bipartite graph that has two types of elements: places and transitions. Place elements are depicted as white circles and transition elements are depicted as rectangles. A place can contain any number of tokens, depicted as black circles.